Merge branch 'master' of github.com:Mintplex-Labs/anything-llm into render

README.md

@@ -87,6 +87,7 @@ AnythingLLM divides your documents into objects called `workspaces`. A Workspace
 - [Fireworks AI  (chat models)](https://fireworks.ai/)
 - [Perplexity (chat models)](https://www.perplexity.ai/)
 - [OpenRouter (chat models)](https://openrouter.ai/)
+- [DeepSeek (chat models)](https://deepseek.com/)
 - [Mistral](https://mistral.ai/)
 - [Groq](https://groq.com/)
 - [Cohere](https://cohere.com/)

collector/package.json

@@ -15,6 +15,7 @@
     "lint": "yarn prettier --ignore-path ../.prettierignore --write ./processSingleFile ./processLink ./utils index.js"
   },
   "dependencies": {
+    "@langchain/community": "^0.2.23",
     "@xenova/transformers": "^2.11.0",
     "bcrypt": "^5.1.0",
     "body-parser": "^1.20.2",
@@ -30,7 +31,6 @@
     "mammoth": "^1.6.0",
     "mbox-parser": "^1.0.1",
     "mime": "^3.0.0",
-    "minimatch": "5.1.0",
     "moment": "^2.29.4",
     "node-html-parser": "^6.1.13",
     "officeparser": "^4.0.5",

collector/utils/extensions/Confluence/ConfluenceLoader/index.js

@@ -13,6 +13,7 @@ class ConfluencePagesLoader {
     limit = 25,
     expand = "body.storage,version",
     personalAccessToken,
+    cloud = true,
   }) {
     this.baseUrl = baseUrl;
     this.spaceKey = spaceKey;
@@ -21,6 +22,7 @@ class ConfluencePagesLoader {
     this.limit = limit;
     this.expand = expand;
     this.personalAccessToken = personalAccessToken;
+    this.cloud = cloud;
   }
 
   get authorizationHeader() {
@@ -74,7 +76,11 @@ class ConfluencePagesLoader {
 
   // https://developer.atlassian.com/cloud/confluence/rest/v2/intro/#auth
   async fetchAllPagesInSpace(start = 0, limit = this.limit) {
-    const url = `${this.baseUrl}/wiki/rest/api/content?spaceKey=${this.spaceKey}&limit=${limit}&start=${start}&expand=${this.expand}`;
+    const url = `${this.baseUrl}${
+      this.cloud ? "/wiki" : ""
+    }/rest/api/content?spaceKey=${
+      this.spaceKey
+    }&limit=${limit}&start=${start}&expand=${this.expand}`;
     const data = await this.fetchConfluenceData(url);
     if (data.size === 0) {
       return [];

collector/utils/extensions/Confluence/index.js

@@ -13,7 +13,13 @@ const { ConfluencePagesLoader } = require("./ConfluenceLoader");
  * @returns
  */
 async function loadConfluence(
-  { baseUrl = null, spaceKey = null, username = null, accessToken = null },
+  {
+    baseUrl = null,
+    spaceKey = null,
+    username = null,
+    accessToken = null,
+    cloud = true,
+  },
   response
 ) {
   if (!baseUrl || !spaceKey || !username || !accessToken) {
@@ -45,6 +51,7 @@ async function loadConfluence(
     spaceKey,
     username,
     accessToken,
+    cloud,
   });
 
   const { docs, error } = await loader
@@ -66,7 +73,7 @@ async function loadConfluence(
     };
   }
   const outFolder = slugify(
-    `confluence-${origin}-${v4().slice(0, 4)}`
+    `confluence-${hostname}-${v4().slice(0, 4)}`
   ).toLowerCase();
   const outFolderPath = path.resolve(documentsFolder, outFolder);
   if (!fs.existsSync(outFolderPath))
@@ -83,7 +90,7 @@ async function loadConfluence(
       description: doc.metadata.title,
       docSource: `${origin} Confluence`,
       chunkSource: generateChunkSource(
-        { doc, baseUrl: origin, spaceKey, accessToken, username },
+        { doc, baseUrl: origin, spaceKey, accessToken, username, cloud },
         response.locals.encryptionWorker
       ),
       published: new Date().toLocaleString(),
@@ -122,6 +129,7 @@ async function fetchConfluencePage({
   spaceKey,
   username,
   accessToken,
+  cloud = true,
 }) {
   if (!pageUrl || !baseUrl || !spaceKey || !username || !accessToken) {
     return {
@@ -154,6 +162,7 @@ async function fetchConfluencePage({
     spaceKey,
     username,
     accessToken,
+    cloud,
   });
 
   const { docs, error } = await loader
@@ -217,7 +226,7 @@ function validBaseUrl(baseUrl) {
  * @returns {string}
  */
 function generateChunkSource(
-  { doc, baseUrl, spaceKey, accessToken, username },
+  { doc, baseUrl, spaceKey, accessToken, username, cloud },
   encryptionWorker
 ) {
   const payload = {
@@ -225,6 +234,7 @@ function generateChunkSource(
     spaceKey,
     token: accessToken,
     username,
+    cloud,
   };
   return `confluence://${doc.metadata.url}?payload=${encryptionWorker.encrypt(
     JSON.stringify(payload)

collector/utils/extensions/RepoLoader/GithubRepo/RepoLoader/index.js

@@ -105,7 +105,7 @@ class GitHubRepoLoader {
     if (!this.ready) throw new Error("[Github Loader]: not in ready state!");
     const {
       GithubRepoLoader: LCGithubLoader,
-    } = require("langchain/document_loaders/web/github");
+    } = require("@langchain/community/document_loaders/web/github");
 
     if (this.accessToken)
       console.log(
@@ -113,17 +113,16 @@ class GitHubRepoLoader {
       );
 
     const loader = new LCGithubLoader(this.repo, {
-      accessToken: this.accessToken,
       branch: this.branch,
       recursive: !!this.accessToken, // Recursive will hit rate limits.
       maxConcurrency: 5,
-      unknown: "ignore",
+      unknown: "warn",
+      accessToken: this.accessToken,
       ignorePaths: this.ignorePaths,
       verbose: true,
     });
 
-    const docs = [];
+    const docs = await loader.load();
-    for await (const doc of loader.loadAsStream()) docs.push(doc);
     return docs;
   }
 

collector/utils/extensions/RepoLoader/GitlabRepo/RepoLoader/index.js

@@ -1,4 +1,4 @@
-const minimatch = require("minimatch");
+const ignore = require("ignore");
 
 /**
  * @typedef {Object} RepoLoaderArgs
@@ -6,6 +6,7 @@ const minimatch = require("minimatch");
  * @property {string} [branch] - The branch to load from (optional).
  * @property {string} [accessToken] - GitLab access token for authentication (optional).
  * @property {string[]} [ignorePaths] - Array of paths to ignore when loading (optional).
+ * @property {boolean} [fetchIssues] - Should issues be fetched (optional).
  */
 
 /**
@@ -33,6 +34,8 @@ class GitLabRepoLoader {
     this.branch = args?.branch;
     this.accessToken = args?.accessToken || null;
     this.ignorePaths = args?.ignorePaths || [];
+    this.ignoreFilter = ignore().add(this.ignorePaths);
+    this.withIssues = args?.fetchIssues || false;
 
     this.projectId = null;
     this.apiBase = "https://gitlab.com";
@@ -123,22 +126,44 @@ class GitLabRepoLoader {
 
     if (this.accessToken)
       console.log(
-        `[Gitlab Loader]: Access token set! Recursive loading enabled!`
+        `[Gitlab Loader]: Access token set! Recursive loading enabled for ${this.repo}!`
       );
 
-    const files = await this.fetchFilesRecursive();
     const docs = [];
 
-    for (const file of files) {
+    console.log(`[Gitlab Loader]: Fetching files.`);
-      if (this.ignorePaths.some((path) => file.path.includes(path))) continue;
+
+    const files = await this.fetchFilesRecursive();
+
+    console.log(`[Gitlab Loader]: Fetched ${files.length} files.`);
+
+    for (const file of files) {
+      if (this.ignoreFilter.ignores(file.path)) continue;
 
-      const content = await this.fetchSingleFileContents(file.path);
-      if (content) {
       docs.push({
-          pageContent: content,
+        pageContent: file.content,
-          metadata: { source: file.path },
+        metadata: {
+          source: file.path,
+          url: `${this.repo}/-/blob/${this.branch}/${file.path}`,
+        },
       });
     }
+
+    if (this.withIssues) {
+      console.log(`[Gitlab Loader]: Fetching issues.`);
+      const issues = await this.fetchIssues();
+      console.log(
+        `[Gitlab Loader]: Fetched ${issues.length} issues with discussions.`
+      );
+      docs.push(
+        ...issues.map((issue) => ({
+          issue,
+          metadata: {
+            source: `issue-${this.repo}-${issue.iid}`,
+            url: issue.web_url,
+          },
+        }))
+      );
     }
 
     return docs;
@@ -160,51 +185,14 @@ class GitLabRepoLoader {
     if (!this.#validGitlabUrl() || !this.projectId) return [];
     await this.#validateAccessToken();
     this.branches = [];
-    let fetching = true;
-    let page = 1;
-    let perPage = 50;
 
-    while (fetching) {
+    const branchesRequestData = {
-      try {
+      endpoint: `/api/v4/projects/${this.projectId}/repository/branches`,
-        const params = new URLSearchParams({
+    };
-          per_page: perPage,
-          page,
-        });
-        const response = await fetch(
-          `${this.apiBase}/api/v4/projects/${
-            this.projectId
-          }/repository/branches?${params.toString()}`,
-          {
-            method: "GET",
-            headers: {
-              Accepts: "application/json",
-              ...(this.accessToken
-                ? { "PRIVATE-TOKEN": this.accessToken }
-                : {}),
-            },
-          }
-        )
-          .then((res) => res.json())
-          .then((branches) => {
-            if (!Array.isArray(branches) || branches.length === 0) {
-              fetching = false;
-              return [];
-            }
-            return branches.map((b) => b.name);
-          })
-          .catch((e) => {
-            console.error(e);
-            fetching = false;
-            return [];
-          });
 
-        this.branches.push(...response);
+    let branchesPage = [];
-        page++;
+    while ((branchesPage = await this.fetchNextPage(branchesRequestData))) {
-      } catch (err) {
+      this.branches.push(...branchesPage.map((branch) => branch.name));
-        console.log(`RepoLoader.getRepoBranches`, err);
-        fetching = false;
-        return [];
-      }
     }
     return this.#branchPrefSort(this.branches);
   }
@@ -215,62 +203,91 @@ class GitLabRepoLoader {
    */
   async fetchFilesRecursive() {
     const files = [];
-    let perPage = 100;
+    const filesRequestData = {
-    let fetching = true;
+      endpoint: `/api/v4/projects/${this.projectId}/repository/tree`,
-    let page = 1;
+      queryParams: {
-
-    while (fetching) {
-      try {
-        const params = new URLSearchParams({
         ref: this.branch,
         recursive: true,
-          per_page: perPage,
+      },
-          page,
+    };
+
+    let filesPage = null;
+    let pagePromises = [];
+    while ((filesPage = await this.fetchNextPage(filesRequestData))) {
+      // Fetch all the files that are not ignored in parallel.
+      pagePromises = filesPage
+        .filter((file) => {
+          if (file.type !== "blob") return false;
+          return !this.ignoreFilter.ignores(file.path);
+        })
+        .map(async (file) => {
+          const content = await this.fetchSingleFileContents(file.path);
+          if (!content) return null;
+          return {
+            path: file.path,
+            content,
+          };
         });
-        const queryUrl = `${this.apiBase}/api/v4/projects/${
-          this.projectId
-        }/repository/tree?${params.toString()}`;
-        const response = await fetch(queryUrl, {
-          method: "GET",
-          headers: this.accessToken
-            ? { "PRIVATE-TOKEN": this.accessToken }
-            : {},
-        });
-        const totalPages = Number(response.headers.get("x-total-pages"));
-        const nextPage = Number(response.headers.get("x-next-page"));
-        const data = await response.json();
 
-        /** @type {FileTreeObject[]} */
+      const pageFiles = await Promise.all(pagePromises);
-        const objects = Array.isArray(data)
-          ? data.filter((item) => item.type === "blob")
-          : []; // only get files, not paths or submodules
 
-        // Apply ignore path rules to found objects. If any rules match it is an invalid file path.
+      files.push(...pageFiles.filter((item) => item !== null));
-        console.log(
+      console.log(`Fetched ${files.length} files.`);
-          `Found ${objects.length} blobs from repo from pg ${page}/${totalPages}`
-        );
-        for (const file of objects) {
-          const isIgnored = this.ignorePaths.some((ignorePattern) =>
-            minimatch(file.path, ignorePattern, { matchBase: true })
-          );
-          if (!isIgnored) files.push(file);
-        }
-
-        if (page === totalPages) {
-          fetching = false;
-          break;
-        }
-
-        page = Number(nextPage);
-      } catch (e) {
-        console.error(`RepoLoader.getRepositoryTree`, e);
-        fetching = false;
-        break;
-      }
     }
+    console.log(`Total files fetched: ${files.length}`);
     return files;
   }
 
+  /**
+   * Fetches all issues from the repository.
+   * @returns {Promise<Issue[]>} An array of issue objects.
+   */
+  async fetchIssues() {
+    const issues = [];
+    const issuesRequestData = {
+      endpoint: `/api/v4/projects/${this.projectId}/issues`,
+    };
+
+    let issuesPage = null;
+    let pagePromises = [];
+    while ((issuesPage = await this.fetchNextPage(issuesRequestData))) {
+      // Fetch all the issues in parallel.
+      pagePromises = issuesPage.map(async (issue) => {
+        const discussionsRequestData = {
+          endpoint: `/api/v4/projects/${this.projectId}/issues/${issue.iid}/discussions`,
+        };
+        let discussionPage = null;
+        const discussions = [];
+
+        while (
+          (discussionPage = await this.fetchNextPage(discussionsRequestData))
+        ) {
+          discussions.push(
+            ...discussionPage.map(({ notes }) =>
+              notes.map(
+                ({ body, author, created_at }) =>
+                  `${author.username} at ${created_at}:
+${body}`
+              )
+            )
+          );
+        }
+        const result = {
+          ...issue,
+          discussions,
+        };
+        return result;
+      });
+
+      const pageIssues = await Promise.all(pagePromises);
+
+      issues.push(...pageIssues);
+      console.log(`Fetched ${issues.length} issues.`);
+    }
+    console.log(`Total issues fetched: ${issues.length}`);
+    return issues;
+  }
+
   /**
    * Fetches the content of a single file from the repository.
    * @param {string} sourceFilePath - The path to the file in the repository.
@@ -301,6 +318,59 @@ class GitLabRepoLoader {
       return null;
     }
   }
+
+  /**
+   * Fetches the next page of data from the API.
+   * @param {Object} requestData - The request data.
+   * @returns {Promise<Array<Object>|null>} The next page of data, or null if no more pages.
+   */
+  async fetchNextPage(requestData) {
+    try {
+      if (requestData.page === -1) return null;
+      if (!requestData.page) requestData.page = 1;
+
+      const { endpoint, perPage = 100, queryParams = {} } = requestData;
+      const params = new URLSearchParams({
+        ...queryParams,
+        per_page: perPage,
+        page: requestData.page,
+      });
+      const url = `${this.apiBase}${endpoint}?${params.toString()}`;
+
+      const response = await fetch(url, {
+        method: "GET",
+        headers: this.accessToken ? { "PRIVATE-TOKEN": this.accessToken } : {},
+      });
+
+      // Rate limits get hit very often if no PAT is provided
+      if (response.status === 401) {
+        console.warn(`Rate limit hit for ${endpoint}. Skipping.`);
+        return null;
+      }
+
+      const totalPages = Number(response.headers.get("x-total-pages"));
+      const data = await response.json();
+      if (!Array.isArray(data)) {
+        console.warn(`Unexpected response format for ${endpoint}:`, data);
+        return [];
+      }
+
+      console.log(
+        `Gitlab RepoLoader: fetched ${endpoint} page ${requestData.page}/${totalPages} with ${data.length} records.`
+      );
+
+      if (totalPages === requestData.page) {
+        requestData.page = -1;
+      } else {
+        requestData.page = Number(response.headers.get("x-next-page"));
+      }
+
+      return data;
+    } catch (e) {
+      console.error(`RepoLoader.fetchNextPage`, e);
+      return null;
+    }
+  }
 }
 
 module.exports = GitLabRepoLoader;

collector/utils/extensions/RepoLoader/GitlabRepo/index.js

@@ -43,13 +43,12 @@ async function loadGitlabRepo(args, response) {
     fs.mkdirSync(outFolderPath, { recursive: true });
 
   for (const doc of docs) {
-    if (!doc.pageContent) continue;
+    if (!doc.metadata || (!doc.pageContent && !doc.issue)) continue;
+    let pageContent = null;
+
     const data = {
       id: v4(),
       url: "gitlab://" + doc.metadata.source,
-      title: doc.metadata.source,
-      docAuthor: repo.author,
-      description: "No description found.",
       docSource: doc.metadata.source,
       chunkSource: generateChunkSource(
         repo,
@@ -57,13 +56,32 @@ async function loadGitlabRepo(args, response) {
         response.locals.encryptionWorker
       ),
       published: new Date().toLocaleString(),
-      wordCount: doc.pageContent.split(" ").length,
-      pageContent: doc.pageContent,
-      token_count_estimate: tokenizeString(doc.pageContent).length,
     };
+
+    if (doc.pageContent) {
+      pageContent = doc.pageContent;
+
+      data.title = doc.metadata.source;
+      data.docAuthor = repo.author;
+      data.description = "No description found.";
+    } else if (doc.issue) {
+      pageContent = issueToMarkdown(doc.issue);
+
+      data.title = `Issue ${doc.issue.iid}: ${doc.issue.title}`;
+      data.docAuthor = doc.issue.author.username;
+      data.description = doc.issue.description;
+    } else {
+      continue;
+    }
+
+    data.wordCount = pageContent.split(" ").length;
+    data.token_count_estimate = tokenizeString(pageContent).length;
+    data.pageContent = pageContent;
+
     console.log(
       `[GitLab Loader]: Saving ${doc.metadata.source} to ${outFolder}`
     );
+
     writeToServerDocuments(
       data,
       `${slugify(doc.metadata.source)}-${data.id}`,
@@ -135,4 +153,93 @@ function generateChunkSource(repo, doc, encryptionWorker) {
   )}`;
 }
 
+function issueToMarkdown(issue) {
+  const metadata = {};
+
+  const userFields = ["author", "assignees", "closed_by"];
+  const userToUsername = ({ username }) => username;
+  for (const userField of userFields) {
+    if (issue[userField]) {
+      if (Array.isArray(issue[userField])) {
+        metadata[userField] = issue[userField].map(userToUsername);
+      } else {
+        metadata[userField] = userToUsername(issue[userField]);
+      }
+    }
+  }
+
+  const singleValueFields = [
+    "web_url",
+    "state",
+    "created_at",
+    "updated_at",
+    "closed_at",
+    "due_date",
+    "type",
+    "merge_request_count",
+    "upvotes",
+    "downvotes",
+    "labels",
+    "has_tasks",
+    "task_status",
+    "confidential",
+    "severity",
+  ];
+
+  for (const singleValueField of singleValueFields) {
+    metadata[singleValueField] = issue[singleValueField];
+  }
+
+  if (issue.milestone) {
+    metadata.milestone = `${issue.milestone.title} (${issue.milestone.id})`;
+  }
+
+  if (issue.time_stats) {
+    const timeFields = ["time_estimate", "total_time_spent"];
+    for (const timeField of timeFields) {
+      const fieldName = `human_${timeField}`;
+      if (issue?.time_stats[fieldName]) {
+        metadata[timeField] = issue.time_stats[fieldName];
+      }
+    }
+  }
+
+  const metadataString = Object.entries(metadata)
+    .map(([name, value]) => {
+      if (!value || value?.length < 1) {
+        return null;
+      }
+      let result = `- ${name.replace("_", " ")}:`;
+
+      if (!Array.isArray(value)) {
+        result += ` ${value}`;
+      } else {
+        result += "\n" + value.map((s) => `  - ${s}`).join("\n");
+      }
+
+      return result;
+    })
+    .filter((item) => item != null)
+    .join("\n");
+
+  let markdown = `# ${issue.title} (${issue.iid})
+
+${issue.description}
+
+## Metadata
+
+${metadataString}`;
+
+  if (issue.discussions.length > 0) {
+    markdown += `
+
+## Activity
+
+${issue.discussions.join("\n\n")}
+`;
+  }
+
+  return markdown;
+}
+
 module.exports = { loadGitlabRepo, fetchGitlabFile };

collector/yarn.lock

@@ -64,6 +64,23 @@
   resolved "https://registry.yarnpkg.com/@huggingface/jinja/-/jinja-0.2.2.tgz#faeb205a9d6995089bef52655ddd8245d3190627"
   integrity sha512-/KPde26khDUIPkTGU82jdtTW9UAuvUTumCAbFs/7giR0SxsvZC4hru51PBvpijH6BVkHcROcvZM/lpy5h1jRRA==
 
+"@langchain/community@^0.2.23":
+  version "0.2.23"
+  resolved "https://registry.yarnpkg.com/@langchain/community/-/community-0.2.23.tgz#20560e107bcc8432c42e499f1b9292d41a3732f2"
+  integrity sha512-p1n/zZ1F+O5l51RzeoUeJyhpzq6Wp11tkqKOj8oThKOQJgLhO7q6iFIvmKThzL7mZCNNuPM5r1OPnU4wO6iF/A==
+  dependencies:
+    "@langchain/core" ">=0.2.16 <0.3.0"
+    "@langchain/openai" ">=0.1.0 <0.3.0"
+    binary-extensions "^2.2.0"
+    expr-eval "^2.0.2"
+    flat "^5.0.2"
+    js-yaml "^4.1.0"
+    langchain "~0.2.3"
+    langsmith "~0.1.30"
+    uuid "^10.0.0"
+    zod "^3.22.3"
+    zod-to-json-schema "^3.22.5"
+
 "@langchain/community@~0.0.47":
   version "0.0.53"
   resolved "https://registry.yarnpkg.com/@langchain/community/-/community-0.0.53.tgz#a9aaedffa0ed2977e8d302d74e9f90a49a6da037"
@@ -78,6 +95,23 @@
     zod "^3.22.3"
     zod-to-json-schema "^3.22.5"
 
+"@langchain/core@>=0.2.11 <0.3.0", "@langchain/core@>=0.2.16 <0.3.0":
+  version "0.2.20"
+  resolved "https://registry.yarnpkg.com/@langchain/core/-/core-0.2.20.tgz#5115781b0a86db3ce4b697e473405892c09621ca"
+  integrity sha512-WPBjrzOj79/yqjloDUIw1GDhuRQfHis07TyyDj+qS81nHh0svSasetKcqAZ3L5JoPcBmEL7rRBtM+OcyC3mLVg==
+  dependencies:
+    ansi-styles "^5.0.0"
+    camelcase "6"
+    decamelize "1.2.0"
+    js-tiktoken "^1.0.12"
+    langsmith "~0.1.39"
+    mustache "^4.2.0"
+    p-queue "^6.6.2"
+    p-retry "4"
+    uuid "^10.0.0"
+    zod "^3.22.4"
+    zod-to-json-schema "^3.22.3"
+
 "@langchain/core@~0.1", "@langchain/core@~0.1.56", "@langchain/core@~0.1.60":
   version "0.1.61"
   resolved "https://registry.yarnpkg.com/@langchain/core/-/core-0.1.61.tgz#9313363e04f1c6981a938b2909c44ce6fceb2736"
@@ -96,6 +130,17 @@
     zod "^3.22.4"
     zod-to-json-schema "^3.22.3"
 
+"@langchain/openai@>=0.1.0 <0.3.0":
+  version "0.2.5"
+  resolved "https://registry.yarnpkg.com/@langchain/openai/-/openai-0.2.5.tgz#e85b983986a7415ea743d4c854bb0674134334d4"
+  integrity sha512-gQXS5VBFyAco0jgSnUVan6fYVSIxlffmDaeDGpXrAmz2nQPgiN/h24KYOt2NOZ1zRheRzRuO/CfRagMhyVUaFA==
+  dependencies:
+    "@langchain/core" ">=0.2.16 <0.3.0"
+    js-tiktoken "^1.0.12"
+    openai "^4.49.1"
+    zod "^3.22.4"
+    zod-to-json-schema "^3.22.3"
+
 "@langchain/openai@~0.0.28":
   version "0.0.28"
   resolved "https://registry.yarnpkg.com/@langchain/openai/-/openai-0.0.28.tgz#afaeec61b44816935db9ae937496c964c81ab571"
@@ -559,13 +604,6 @@ brace-expansion@^1.1.7:
     balanced-match "^1.0.0"
     concat-map "0.0.1"
 
-brace-expansion@^2.0.1:
-  version "2.0.1"
-  resolved "https://registry.yarnpkg.com/brace-expansion/-/brace-expansion-2.0.1.tgz#1edc459e0f0c548486ecf9fc99f2221364b9a0ae"
-  integrity sha512-XnAIvQ8eM+kC6aULx6wuQiwVsnzsi9d3WxzV3FpWTGA19F621kwdbsAcFKXgKUHZWsy+mY6iL1sHTxWEFCytDA==
-  dependencies:
-    balanced-match "^1.0.0"
-
 braces@~3.0.2:
   version "3.0.2"
   resolved "https://registry.yarnpkg.com/braces/-/braces-3.0.2.tgz#3454e1a462ee8d599e236df336cd9ea4f8afe107"
@@ -1769,6 +1807,13 @@ js-tiktoken@^1.0.11, js-tiktoken@^1.0.7, js-tiktoken@^1.0.8:
   dependencies:
     base64-js "^1.5.1"
 
+js-tiktoken@^1.0.12:
+  version "1.0.12"
+  resolved "https://registry.yarnpkg.com/js-tiktoken/-/js-tiktoken-1.0.12.tgz#af0f5cf58e5e7318240d050c8413234019424211"
+  integrity sha512-L7wURW1fH9Qaext0VzaUDpFGVQgjkdE3Dgsy9/+yXyGEpBKnylTd0mU0bfbNkKDlXRb6TEsZkwuflu1B8uQbJQ==
+  dependencies:
+    base64-js "^1.5.1"
+
 js-tokens@^4.0.0:
   version "4.0.0"
   resolved "https://registry.yarnpkg.com/js-tokens/-/js-tokens-4.0.0.tgz#19203fb59991df98e3a287050d4647cdeaf32499"
@@ -1844,6 +1889,28 @@ langchain@0.1.36:
     zod "^3.22.4"
     zod-to-json-schema "^3.22.3"
 
+langchain@~0.2.3:
+  version "0.2.12"
+  resolved "https://registry.yarnpkg.com/langchain/-/langchain-0.2.12.tgz#3fac0b9519a070689b6dd679d5854abc57824dcf"
+  integrity sha512-ZHtJrHUpridZ7IQu7N/wAQ6iMAAO7VLzkupHqKP79S6p+alrPbn1BjRnh+PeGm92YiY5DafTCuvchmujxx7bCQ==
+  dependencies:
+    "@langchain/core" ">=0.2.11 <0.3.0"
+    "@langchain/openai" ">=0.1.0 <0.3.0"
+    "@langchain/textsplitters" "~0.0.0"
+    binary-extensions "^2.2.0"
+    js-tiktoken "^1.0.12"
+    js-yaml "^4.1.0"
+    jsonpointer "^5.0.1"
+    langchainhub "~0.0.8"
+    langsmith "~0.1.30"
+    ml-distance "^4.0.0"
+    openapi-types "^12.1.3"
+    p-retry "4"
+    uuid "^10.0.0"
+    yaml "^2.2.1"
+    zod "^3.22.4"
+    zod-to-json-schema "^3.22.3"
+
 langchainhub@~0.0.8:
   version "0.0.8"
   resolved "https://registry.yarnpkg.com/langchainhub/-/langchainhub-0.0.8.tgz#fd4b96dc795e22e36c1a20bad31b61b0c33d3110"
@@ -1860,6 +1927,18 @@ langsmith@~0.1.1, langsmith@~0.1.7:
     p-retry "4"
     uuid "^9.0.0"
 
+langsmith@~0.1.30, langsmith@~0.1.39:
+  version "0.1.40"
+  resolved "https://registry.yarnpkg.com/langsmith/-/langsmith-0.1.40.tgz#9708889386a5b9d0eb43dd3a9eba93513b57101d"
+  integrity sha512-11E2WLbh/+41+Qc0w8fJJTC/iz91BA+zXRMX/Wz0KSstnfzIPBoiWa++Kp2X8yCIDNywWWLJhy/B8gYzm7VKig==
+  dependencies:
+    "@types/uuid" "^9.0.1"
+    commander "^10.0.1"
+    p-queue "^6.6.2"
+    p-retry "4"
+    semver "^7.6.3"
+    uuid "^9.0.0"
+
 leac@^0.6.0:
   version "0.6.0"
   resolved "https://registry.yarnpkg.com/leac/-/leac-0.6.0.tgz#dcf136e382e666bd2475f44a1096061b70dc0912"
@@ -2082,13 +2161,6 @@ mimic-response@^3.1.0:
   resolved "https://registry.yarnpkg.com/mimic-response/-/mimic-response-3.1.0.tgz#2d1d59af9c1b129815accc2c46a022a5ce1fa3c9"
   integrity sha512-z0yWI+4FDrrweS8Zmt4Ej5HdJmky15+L2e6Wgn3+iK5fWzb6T3fhNFq2+MeTRb064c6Wr4N/wv0DzQTjNzHNGQ==
 
-minimatch@5.1.0:
-  version "5.1.0"
-  resolved "https://registry.yarnpkg.com/minimatch/-/minimatch-5.1.0.tgz#1717b464f4971b144f6aabe8f2d0b8e4511e09c7"
-  integrity sha512-9TPBGGak4nHfGZsPBohm9AWg6NoT7QTCehS3BIJABslyZbzxfV78QM2Y6+i741OPZIafFAaiiEMh5OyIrJPgtg==
-  dependencies:
-    brace-expansion "^2.0.1"
-
 minimatch@^3.1.1, minimatch@^3.1.2:
   version "3.1.2"
   resolved "https://registry.yarnpkg.com/minimatch/-/minimatch-3.1.2.tgz#19cd194bfd3e428f049a70817c038d89ab4be35b"
@@ -2417,6 +2489,19 @@ openai@^4.32.1:
     node-fetch "^2.6.7"
     web-streams-polyfill "^3.2.1"
 
+openai@^4.49.1:
+  version "4.54.0"
+  resolved "https://registry.yarnpkg.com/openai/-/openai-4.54.0.tgz#eeb209c6892b997e524181b6ddb7e27bf4d09389"
+  integrity sha512-e/12BdtTtj+tXs7iHm+Dm7H7WjEWnw7O52B2wSfCQ6lD5F6cvjzo7cANXy5TJ1Q3/qc8YRPT5wBTTFtP5sBp1g==
+  dependencies:
+    "@types/node" "^18.11.18"
+    "@types/node-fetch" "^2.6.4"
+    abort-controller "^3.0.0"
+    agentkeepalive "^4.2.1"
+    form-data-encoder "1.7.2"
+    formdata-node "^4.3.2"
+    node-fetch "^2.6.7"
+
 openapi-types@^12.1.3:
   version "12.1.3"
   resolved "https://registry.yarnpkg.com/openapi-types/-/openapi-types-12.1.3.tgz#471995eb26c4b97b7bd356aacf7b91b73e777dd3"
@@ -2863,6 +2948,11 @@ semver@^7.3.5, semver@^7.5.4:
   dependencies:
     lru-cache "^6.0.0"
 
+semver@^7.6.3:
+  version "7.6.3"
+  resolved "https://registry.yarnpkg.com/semver/-/semver-7.6.3.tgz#980f7b5550bc175fb4dc09403085627f9eb33143"
+  integrity sha512-oVekP1cKtI+CTDvHWYFUcMtsK/00wmAEfyqKfNdARm8u1wNVhSgaX7A8d4UuIlUI5e84iEwOhs7ZPYRmzU9U6A==
+
 semver@~7.0.0:
   version "7.0.0"
   resolved "https://registry.yarnpkg.com/semver/-/semver-7.0.0.tgz#5f3ca35761e47e05b206c6daff2cf814f0316b8e"
@@ -3336,6 +3426,11 @@ utils-merge@1.0.1:
   resolved "https://registry.yarnpkg.com/utils-merge/-/utils-merge-1.0.1.tgz#9f95710f50a267947b2ccc124741c1028427e713"
   integrity sha512-pMZTvIkT1d+TFGvDOqodOclx0QWkkgi6Tdoa8gC8ffGAAqz9pzPTZWAybbsHHoED/ztMtkv/VoYTYyShUn81hA==
 
+uuid@^10.0.0:
+  version "10.0.0"
+  resolved "https://registry.yarnpkg.com/uuid/-/uuid-10.0.0.tgz#5a95aa454e6e002725c79055fd42aaba30ca6294"
+  integrity sha512-8XkAphELsDnEGrDxUOHB3RGvXz6TeuYSGEZBOjtTtPm2lwhGBjLgOzLHB63IUWfBpNucQjND6d3AOudO+H3RWQ==
+
 uuid@^9.0.0:
   version "9.0.1"
   resolved "https://registry.yarnpkg.com/uuid/-/uuid-9.0.1.tgz#e188d4c8853cc722220392c424cd637f32293f30"

frontend/src/components/LLMSelection/DeepSeekOptions/index.jsx

@@ -0,0 +1,100 @@
+import { useState, useEffect } from "react";
+import System from "@/models/system";
+
+export default function DeepSeekOptions({ settings }) {
+  const [inputValue, setInputValue] = useState(settings?.DeepSeekApiKey);
+  const [deepSeekApiKey, setDeepSeekApiKey] = useState(
+    settings?.DeepSeekApiKey
+  );
+
+  return (
+    <div className="flex gap-[36px] mt-1.5">
+      <div className="flex flex-col w-60">
+        <label className="text-white text-sm font-semibold block mb-3">
+          API Key
+        </label>
+        <input
+          type="password"
+          name="DeepSeekApiKey"
+          className="bg-zinc-900 text-white placeholder:text-white/20 text-sm rounded-lg focus:outline-primary-button active:outline-primary-button outline-none block w-full p-2.5"
+          placeholder="DeepSeek API Key"
+          defaultValue={settings?.DeepSeekApiKey ? "*".repeat(20) : ""}
+          required={true}
+          autoComplete="off"
+          spellCheck={false}
+          onChange={(e) => setInputValue(e.target.value)}
+          onBlur={() => setDeepSeekApiKey(inputValue)}
+        />
+      </div>
+      {!settings?.credentialsOnly && (
+        <DeepSeekModelSelection settings={settings} apiKey={deepSeekApiKey} />
+      )}
+    </div>
+  );
+}
+
+function DeepSeekModelSelection({ apiKey, settings }) {
+  const [models, setModels] = useState([]);
+  const [loading, setLoading] = useState(true);
+
+  useEffect(() => {
+    async function findCustomModels() {
+      if (!apiKey) {
+        setModels([]);
+        setLoading(true);
+        return;
+      }
+
+      setLoading(true);
+      const { models } = await System.customModels(
+        "deepseek",
+        typeof apiKey === "boolean" ? null : apiKey
+      );
+      setModels(models || []);
+      setLoading(false);
+    }
+    findCustomModels();
+  }, [apiKey]);
+
+  if (loading) {
+    return (
+      <div className="flex flex-col w-60">
+        <label className="text-white text-sm font-semibold block mb-3">
+          Chat Model Selection
+        </label>
+        <select
+          name="DeepSeekModelPref"
+          disabled={true}
+          className="bg-zinc-900 border-gray-500 text-white text-sm rounded-lg block w-full p-2.5"
+        >
+          <option disabled={true} selected={true}>
+            -- loading available models --
+          </option>
+        </select>
+      </div>
+    );
+  }
+
+  return (
+    <div className="flex flex-col w-60">
+      <label className="text-white text-sm font-semibold block mb-3">
+        Chat Model Selection
+      </label>
+      <select
+        name="DeepSeekModelPref"
+        required={true}
+        className="bg-zinc-900 border-gray-500 text-white text-sm rounded-lg block w-full p-2.5"
+      >
+        {models.map((model) => (
+          <option
+            key={model.id}
+            value={model.id}
+            selected={settings?.DeepSeekModelPref === model.id}
+          >
+            {model.name}
+          </option>
+        ))}
+      </select>
+    </div>
+  );
+}

frontend/src/components/Modals/ManageWorkspace/DataConnectors/Connectors/Confluence/index.jsx

@@ -26,6 +26,7 @@ export default function ConfluenceOptions() {
         spaceKey: form.get("spaceKey"),
         username: form.get("username"),
         accessToken: form.get("accessToken"),
+        cloud: form.get("isCloud") === "true",
       });
 
       if (!!error) {
@@ -54,6 +55,31 @@ export default function ConfluenceOptions() {
         <form className="w-full" onSubmit={handleSubmit}>
           <div className="w-full flex flex-col py-2">
             <div className="w-full flex flex-col gap-4">
+              <div className="flex flex-col pr-10">
+                <div className="flex flex-col gap-y-1 mb-4">
+                  <label className="text-white text-sm font-bold flex gap-x-2 items-center">
+                    <p className="font-bold text-white">
+                      Confluence deployment type
+                    </p>
+                  </label>
+                  <p className="text-xs font-normal text-white/50">
+                    Determine if your Confluence instance is hosted on Atlassian
+                    cloud or self-hosted.
+                  </p>
+                </div>
+                <select
+                  name="isCloud"
+                  className="bg-zinc-900 text-white placeholder:text-white/20 text-sm rounded-lg focus:outline-primary-button active:outline-primary-button outline-none block w-full p-2.5"
+                  required={true}
+                  autoComplete="off"
+                  spellCheck={false}
+                  defaultValue="true"
+                >
+                  <option value="true">Atlassian Cloud</option>
+                  <option value="false">Self-hosted</option>
+                </select>
+              </div>
+
               <div className="flex flex-col pr-10">
                 <div className="flex flex-col gap-y-1 mb-4">
                   <label className="text-white text-sm font-bold flex gap-x-2 items-center">
@@ -103,7 +129,7 @@ export default function ConfluenceOptions() {
                   </p>
                 </div>
                 <input
-                  type="email"
+                  type="text"
                   name="username"
                   className="bg-zinc-900 text-white placeholder:text-white/20 text-sm rounded-lg focus:outline-primary-button active:outline-primary-button outline-none block w-full p-2.5"
                   placeholder="jdoe@example.com"

frontend/src/components/Modals/ManageWorkspace/DataConnectors/Connectors/Gitlab/index.jsx

@@ -34,6 +34,7 @@ export default function GitlabOptions() {
         accessToken: form.get("accessToken"),
         branch: form.get("branch"),
         ignorePaths: ignores,
+        fetchIssues: form.get("fetchIssues"),
       });
 
       if (!!error) {
@@ -112,6 +113,30 @@ export default function GitlabOptions() {
                   onBlur={() => setSettings({ ...settings, accessToken })}
                 />
               </div>
+              <div className="flex flex-col pr-10">
+                <div className="flex flex-col gap-y-1 mb-4">
+                  <label className="text-white font-bold text-sm flex gap-x-2 items-center">
+                    <p className="font-bold text-white">Settings</p>{" "}
+                  </label>
+                  <p className="text-xs font-normal text-white/50">
+                    Select additional entities to fetch from the GitLab API.
+                  </p>
+                </div>
+                <div className="flex items-center gap-x-2">
+                  <label className="relative inline-flex cursor-pointer items-center">
+                    <input
+                      type="checkbox"
+                      name="fetchIssues"
+                      value={true}
+                      className="peer sr-only"
+                    />
+                    <div className="pointer-events-none peer h-6 w-11 rounded-full bg-stone-400 after:absolute after:left-[2px] after:top-[2px] after:h-5 after:w-5 after:rounded-full after:shadow-xl after:border after:border-gray-600 after:bg-white after:box-shadow-md after:transition-all after:content-[''] peer-checked:bg-lime-300 peer-checked:after:translate-x-full peer-checked:after:border-white peer-focus:outline-none peer-focus:ring-4 peer-focus:ring-blue-800"></div>
+                    <span className="ml-3 text-sm font-medium text-white">
+                      Fetch Issues as Documents
+                    </span>
+                  </label>
+                </div>
+              </div>
               <GitLabBranchSelection
                 repo={settings.repo}
                 accessToken={settings.accessToken}

frontend/src/models/dataConnector.js

@@ -64,11 +64,23 @@ const DataConnector = {
           return { branches: [], error: e.message };
         });
     },
-    collect: async function ({ repo, accessToken, branch, ignorePaths = [] }) {
+    collect: async function ({
+      repo,
+      accessToken,
+      branch,
+      ignorePaths = [],
+      fetchIssues = false,
+    }) {
       return await fetch(`${API_BASE}/ext/gitlab/repo`, {
         method: "POST",
         headers: baseHeaders(),
-        body: JSON.stringify({ repo, accessToken, branch, ignorePaths }),
+        body: JSON.stringify({
+          repo,
+          accessToken,
+          branch,
+          ignorePaths,
+          fetchIssues,
+        }),
       })
         .then((res) => res.json())
         .then((res) => {
@@ -119,7 +131,13 @@ const DataConnector = {
   },
 
   confluence: {
-    collect: async function ({ baseUrl, spaceKey, username, accessToken }) {
+    collect: async function ({
+      baseUrl,
+      spaceKey,
+      username,
+      accessToken,
+      cloud,
+    }) {
       return await fetch(`${API_BASE}/ext/confluence`, {
         method: "POST",
         headers: baseHeaders(),
@@ -128,6 +146,7 @@ const DataConnector = {
           spaceKey,
           username,
           accessToken,
+          cloud,
         }),
       })
         .then((res) => res.json())

frontend/src/pages/GeneralSettings/LLMPreference/index.jsx

@@ -24,6 +24,7 @@ import TextGenWebUILogo from "@/media/llmprovider/text-generation-webui.png";
 import CohereLogo from "@/media/llmprovider/cohere.png";
 import LiteLLMLogo from "@/media/llmprovider/litellm.png";
 import AWSBedrockLogo from "@/media/llmprovider/bedrock.png";
+import DeepSeekLogo from "@/media/llmprovider/deepseek.png";
 
 import PreLoader from "@/components/Preloader";
 import OpenAiOptions from "@/components/LLMSelection/OpenAiOptions";
@@ -46,6 +47,7 @@ import KoboldCPPOptions from "@/components/LLMSelection/KoboldCPPOptions";
 import TextGenWebUIOptions from "@/components/LLMSelection/TextGenWebUIOptions";
 import LiteLLMOptions from "@/components/LLMSelection/LiteLLMOptions";
 import AWSBedrockLLMOptions from "@/components/LLMSelection/AwsBedrockLLMOptions";
+import DeepSeekOptions from "@/components/LLMSelection/DeepSeekOptions";
 
 import LLMItem from "@/components/LLMSelection/LLMItem";
 import { CaretUpDown, MagnifyingGlass, X } from "@phosphor-icons/react";
@@ -209,6 +211,14 @@ export const AVAILABLE_LLM_PROVIDERS = [
     description: "Run LiteLLM's OpenAI compatible proxy for various LLMs.",
     requiredConfig: ["LiteLLMBasePath"],
   },
+  {
+    name: "DeepSeek",
+    value: "deepseek",
+    logo: DeepSeekLogo,
+    options: (settings) => <DeepSeekOptions settings={settings} />,
+    description: "Run DeepSeek's powerful LLMs.",
+    requiredConfig: ["DeepSeekApiKey"],
+  },
   {
     name: "Generic OpenAI",
     value: "generic-openai",

frontend/src/pages/OnboardingFlow/Steps/DataHandling/index.jsx

@@ -20,6 +20,7 @@ import KoboldCPPLogo from "@/media/llmprovider/koboldcpp.png";
 import TextGenWebUILogo from "@/media/llmprovider/text-generation-webui.png";
 import LiteLLMLogo from "@/media/llmprovider/litellm.png";
 import AWSBedrockLogo from "@/media/llmprovider/bedrock.png";
+import DeepSeekLogo from "@/media/llmprovider/deepseek.png";
 
 import CohereLogo from "@/media/llmprovider/cohere.png";
 import ZillizLogo from "@/media/vectordbs/zilliz.png";
@@ -196,6 +197,11 @@ export const LLM_SELECTION_PRIVACY = {
     ],
     logo: AWSBedrockLogo,
   },
+  deepseek: {
+    name: "DeepSeek",
+    description: ["Your model and chat contents are visible to DeepSeek"],
+    logo: DeepSeekLogo,
+  },
 };
 
 export const VECTOR_DB_PRIVACY = {

frontend/src/pages/OnboardingFlow/Steps/LLMPreference/index.jsx

@@ -19,6 +19,7 @@ import KoboldCPPLogo from "@/media/llmprovider/koboldcpp.png";
 import TextGenWebUILogo from "@/media/llmprovider/text-generation-webui.png";
 import LiteLLMLogo from "@/media/llmprovider/litellm.png";
 import AWSBedrockLogo from "@/media/llmprovider/bedrock.png";
+import DeepSeekLogo from "@/media/llmprovider/deepseek.png";
 
 import CohereLogo from "@/media/llmprovider/cohere.png";
 import OpenAiOptions from "@/components/LLMSelection/OpenAiOptions";
@@ -41,6 +42,7 @@ import KoboldCPPOptions from "@/components/LLMSelection/KoboldCPPOptions";
 import TextGenWebUIOptions from "@/components/LLMSelection/TextGenWebUIOptions";
 import LiteLLMOptions from "@/components/LLMSelection/LiteLLMOptions";
 import AWSBedrockLLMOptions from "@/components/LLMSelection/AwsBedrockLLMOptions";
+import DeepSeekOptions from "@/components/LLMSelection/DeepSeekOptions";
 
 import LLMItem from "@/components/LLMSelection/LLMItem";
 import System from "@/models/system";
@@ -184,6 +186,13 @@ const LLMS = [
     options: (settings) => <LiteLLMOptions settings={settings} />,
     description: "Run LiteLLM's OpenAI compatible proxy for various LLMs.",
   },
+  {
+    name: "DeepSeek",
+    value: "deepseek",
+    logo: DeepSeekLogo,
+    options: (settings) => <DeepSeekOptions settings={settings} />,
+    description: "Run DeepSeek's powerful LLMs.",
+  },
   {
     name: "Generic OpenAI",
     value: "generic-openai",

frontend/src/pages/WorkspaceSettings/AgentConfig/AgentLLMSelection/index.jsx

@@ -23,6 +23,7 @@ const ENABLED_PROVIDERS = [
   "generic-openai",
   "bedrock",
   "fireworksai",
+  "deepseek",
   // TODO: More agent support.
   // "cohere",         // Has tool calling and will need to build explicit support
   // "huggingface"     // Can be done but already has issues with no-chat templated. Needs to be tested.

server/models/systemSettings.js

@@ -511,6 +511,10 @@ const SystemSettings = {
 
       // VoyageAi API Keys
       VoyageAiApiKey: !!process.env.VOYAGEAI_API_KEY,
+
+      // DeepSeek API Keys
+      DeepSeekApiKey: !!process.env.DEEPSEEK_API_KEY,
+      DeepSeekModelPref: process.env.DEEPSEEK_MODEL_PREF,
     };
   },
 

server/utils/AiProviders/deepseek/index.js

@@ -0,0 +1,127 @@
+const { NativeEmbedder } = require("../../EmbeddingEngines/native");
+const {
+  handleDefaultStreamResponseV2,
+} = require("../../helpers/chat/responses");
+const { MODEL_MAP } = require("../modelMap");
+
+class DeepSeekLLM {
+  constructor(embedder = null, modelPreference = null) {
+    if (!process.env.DEEPSEEK_API_KEY)
+      throw new Error("No DeepSeek API key was set.");
+    const { OpenAI: OpenAIApi } = require("openai");
+
+    this.openai = new OpenAIApi({
+      apiKey: process.env.DEEPSEEK_API_KEY,
+      baseURL: "https://api.deepseek.com/v1",
+    });
+    this.model =
+      modelPreference || process.env.DEEPSEEK_MODEL_PREF || "deepseek-chat";
+    this.limits = {
+      history: this.promptWindowLimit() * 0.15,
+      system: this.promptWindowLimit() * 0.15,
+      user: this.promptWindowLimit() * 0.7,
+    };
+
+    this.embedder = embedder ?? new NativeEmbedder();
+    this.defaultTemp = 0.7;
+  }
+
+  #appendContext(contextTexts = []) {
+    if (!contextTexts || !contextTexts.length) return "";
+    return (
+      "\nContext:\n" +
+      contextTexts
+        .map((text, i) => {
+          return `[CONTEXT ${i}]:\n${text}\n[END CONTEXT ${i}]\n\n`;
+        })
+        .join("")
+    );
+  }
+
+  streamingEnabled() {
+    return "streamGetChatCompletion" in this;
+  }
+
+  static promptWindowLimit(modelName) {
+    return MODEL_MAP.deepseek[modelName] ?? 8192;
+  }
+
+  promptWindowLimit() {
+    return MODEL_MAP.deepseek[this.model] ?? 8192;
+  }
+
+  async isValidChatCompletionModel(modelName = "") {
+    const models = await this.openai.models.list().catch(() => ({ data: [] }));
+    return models.data.some((model) => model.id === modelName);
+  }
+
+  constructPrompt({
+    systemPrompt = "",
+    contextTexts = [],
+    chatHistory = [],
+    userPrompt = "",
+  }) {
+    const prompt = {
+      role: "system",
+      content: `${systemPrompt}${this.#appendContext(contextTexts)}`,
+    };
+    return [prompt, ...chatHistory, { role: "user", content: userPrompt }];
+  }
+
+  async getChatCompletion(messages = null, { temperature = 0.7 }) {
+    if (!(await this.isValidChatCompletionModel(this.model)))
+      throw new Error(
+        `DeepSeek chat: ${this.model} is not valid for chat completion!`
+      );
+
+    const result = await this.openai.chat.completions
+      .create({
+        model: this.model,
+        messages,
+        temperature,
+      })
+      .catch((e) => {
+        throw new Error(e.message);
+      });
+
+    if (!result.hasOwnProperty("choices") || result.choices.length === 0)
+      return null;
+    return result.choices[0].message.content;
+  }
+
+  async streamGetChatCompletion(messages = null, { temperature = 0.7 }) {
+    if (!(await this.isValidChatCompletionModel(this.model)))
+      throw new Error(
+        `DeepSeek chat: ${this.model} is not valid for chat completion!`
+      );
+
+    const streamRequest = await this.openai.chat.completions.create({
+      model: this.model,
+      stream: true,
+      messages,
+      temperature,
+    });
+    return streamRequest;
+  }
+
+  handleStream(response, stream, responseProps) {
+    return handleDefaultStreamResponseV2(response, stream, responseProps);
+  }
+
+  async embedTextInput(textInput) {
+    return await this.embedder.embedTextInput(textInput);
+  }
+  async embedChunks(textChunks = []) {
+    return await this.embedder.embedChunks(textChunks);
+  }
+
+  async compressMessages(promptArgs = {}, rawHistory = []) {
+    const { messageArrayCompressor } = require("../../helpers/chat");
+    const messageArray = this.constructPrompt(promptArgs);
+    return await messageArrayCompressor(this, messageArray, rawHistory);
+  }
+}
+
+module.exports = {
+  DeepSeekLLM,
+};

server/utils/AiProviders/modelMap.js

@@ -53,6 +53,10 @@ const MODEL_MAP = {
     "gpt-4": 8_192,
     "gpt-4-32k": 32_000,
   },
+  deepseek: {
+    "deepseek-chat": 128_000,
+    "deepseek-coder": 128_000,
+  },
 };
 
 module.exports = { MODEL_MAP };

server/utils/agents/aibitat/index.js

@@ -783,6 +783,8 @@ ${this.getHistory({ to: route.to })
         return new Providers.AWSBedrockProvider({});
       case "fireworksai":
         return new Providers.FireworksAIProvider({ model: config.model });
+      case "deepseek":
+        return new Providers.DeepSeekProvider({ model: config.model });
 
       default:
         throw new Error(

server/utils/agents/aibitat/providers/ai-provider.js

@@ -174,6 +174,14 @@ class Provider {
           apiKey: process.env.TEXT_GEN_WEB_UI_API_KEY ?? "not-used",
           ...config,
         });
+      case "deepseek":
+        return new ChatOpenAI({
+          configuration: {
+            baseURL: "https://api.deepseek.com/v1",
+          },
+          apiKey: process.env.DEEPSEEK_API_KEY ?? null,
+          ...config,
+        });
       default:
         throw new Error(`Unsupported provider ${provider} for this task.`);
     }

server/utils/agents/aibitat/providers/deepseek.js

@@ -0,0 +1,118 @@
+const OpenAI = require("openai");
+const Provider = require("./ai-provider.js");
+const InheritMultiple = require("./helpers/classes.js");
+const UnTooled = require("./helpers/untooled.js");
+const { toValidNumber } = require("../../../http/index.js");
+
+class DeepSeekProvider extends InheritMultiple([Provider, UnTooled]) {
+  model;
+
+  constructor(config = {}) {
+    super();
+    const { model = "deepseek-chat" } = config;
+    const client = new OpenAI({
+      baseURL: "https://api.deepseek.com/v1",
+      apiKey: process.env.DEEPSEEK_API_KEY ?? null,
+      maxRetries: 3,
+    });
+
+    this._client = client;
+    this.model = model;
+    this.verbose = true;
+    this.maxTokens = process.env.DEEPSEEK_MAX_TOKENS
+      ? toValidNumber(process.env.DEEPSEEK_MAX_TOKENS, 1024)
+      : 1024;
+  }
+
+  get client() {
+    return this._client;
+  }
+
+  async #handleFunctionCallChat({ messages = [] }) {
+    return await this.client.chat.completions
+      .create({
+        model: this.model,
+        temperature: 0,
+        messages,
+        max_tokens: this.maxTokens,
+      })
+      .then((result) => {
+        if (!result.hasOwnProperty("choices"))
+          throw new Error("DeepSeek chat: No results!");
+        if (result.choices.length === 0)
+          throw new Error("DeepSeek chat: No results length!");
+        return result.choices[0].message.content;
+      })
+      .catch((_) => {
+        return null;
+      });
+  }
+
+  /**
+   * Create a completion based on the received messages.
+   *
+   * @param messages A list of messages to send to the API.
+   * @param functions
+   * @returns The completion.
+   */
+  async complete(messages, functions = null) {
+    try {
+      let completion;
+      if (functions.length > 0) {
+        const { toolCall, text } = await this.functionCall(
+          messages,
+          functions,
+          this.#handleFunctionCallChat.bind(this)
+        );
+
+        if (toolCall !== null) {
+          this.providerLog(`Valid tool call found - running ${toolCall.name}.`);
+          this.deduplicator.trackRun(toolCall.name, toolCall.arguments);
+          return {
+            result: null,
+            functionCall: {
+              name: toolCall.name,
+              arguments: toolCall.arguments,
+            },
+            cost: 0,
+          };
+        }
+        completion = { content: text };
+      }
+
+      if (!completion?.content) {
+        this.providerLog(
+          "Will assume chat completion without tool call inputs."
+        );
+        const response = await this.client.chat.completions.create({
+          model: this.model,
+          messages: this.cleanMsgs(messages),
+        });
+        completion = response.choices[0].message;
+      }
+
+      // The UnTooled class inherited Deduplicator is mostly useful to prevent the agent
+      // from calling the exact same function over and over in a loop within a single chat exchange
+      // _but_ we should enable it to call previously used tools in a new chat interaction.
+      this.deduplicator.reset("runs");
+      return {
+        result: completion.content,
+        cost: 0,
+      };
+    } catch (error) {
+      throw error;
+    }
+  }
+
+  /**
+   * Get the cost of the completion.
+   *
+   * @param _usage The completion to get the cost for.
+   * @returns The cost of the completion.
+   */
+  getCost(_usage) {
+    return 0;
+  }
+}
+
+module.exports = DeepSeekProvider;

server/utils/agents/aibitat/providers/index.js

@@ -14,6 +14,7 @@ const PerplexityProvider = require("./perplexity.js");
 const TextWebGenUiProvider = require("./textgenwebui.js");
 const AWSBedrockProvider = require("./bedrock.js");
 const FireworksAIProvider = require("./fireworksai.js");
+const DeepSeekProvider = require("./deepseek.js");
 
 module.exports = {
   OpenAIProvider,
@@ -28,6 +29,7 @@ module.exports = {
   OpenRouterProvider,
   MistralProvider,
   GenericOpenAiProvider,
+  DeepSeekProvider,
   PerplexityProvider,
   TextWebGenUiProvider,
   AWSBedrockProvider,

server/utils/agents/index.js

@@ -162,6 +162,10 @@ class AgentHandler {
             "FireworksAI API Key must be provided to use agents."
           );
         break;
+      case "deepseek":
+        if (!process.env.DEEPSEEK_API_KEY)
+          throw new Error("DeepSeek API Key must be provided to use agents.");
+        break;
 
       default:
         throw new Error(
@@ -206,6 +210,8 @@ class AgentHandler {
         return null;
       case "fireworksai":
         return null;
+      case "deepseek":
+        return "deepseek-chat";
       default:
         return "unknown";
     }

server/utils/helpers/customModels.js

@@ -18,6 +18,7 @@ const SUPPORT_CUSTOM_MODELS = [
   "litellm",
   "elevenlabs-tts",
   "groq",
+  "deepseek",
 ];
 
 async function getCustomModels(provider = "", apiKey = null, basePath = null) {
@@ -53,6 +54,8 @@ async function getCustomModels(provider = "", apiKey = null, basePath = null) {
       return await getElevenLabsModels(apiKey);
     case "groq":
       return await getGroqAiModels(apiKey);
+    case "deepseek":
+      return await getDeepSeekModels(apiKey);
     default:
       return { models: [], error: "Invalid provider for custom models" };
   }
@@ -419,6 +422,31 @@ async function getElevenLabsModels(apiKey = null) {
   return { models, error: null };
 }
 
+async function getDeepSeekModels(apiKey = null) {
+  const { OpenAI: OpenAIApi } = require("openai");
+  const openai = new OpenAIApi({
+    apiKey: apiKey || process.env.DEEPSEEK_API_KEY,
+    baseURL: "https://api.deepseek.com/v1",
+  });
+  const models = await openai.models
+    .list()
+    .then((results) => results.data)
+    .then((models) =>
+      models.map((model) => ({
+        id: model.id,
+        name: model.id,
+        organization: model.owned_by,
+      }))
+    )
+    .catch((e) => {
+      console.error(`DeepSeek:listModels`, e.message);
+      return [];
+    });
+
+  if (models.length > 0 && !!apiKey) process.env.DEEPSEEK_API_KEY = apiKey;
+  return { models, error: null };
+}
+
 module.exports = {
   getCustomModels,
 };

server/utils/helpers/index.js

@@ -159,6 +159,9 @@ function getLLMProvider({ provider = null, model = null } = {}) {
     case "bedrock":
       const { AWSBedrockLLM } = require("../AiProviders/bedrock");
       return new AWSBedrockLLM(embedder, model);
+    case "deepseek":
+      const { DeepSeekLLM } = require("../AiProviders/deepseek");
+      return new DeepSeekLLM(embedder, model);
     default:
       throw new Error(
         `ENV: No valid LLM_PROVIDER value found in environment! Using ${process.env.LLM_PROVIDER}`

server/utils/helpers/updateENV.js

@@ -501,6 +501,16 @@ const KEY_MAPPING = {
     envKey: "TTS_PIPER_VOICE_MODEL",
     checks: [],
   },
+
+  // DeepSeek Options
+  DeepSeekApiKey: {
+    envKey: "DEEPSEEK_API_KEY",
+    checks: [isNotEmpty],
+  },
+  DeepSeekModelPref: {
+    envKey: "DEEPSEEK_MODEL_PREF",
+    checks: [isNotEmpty],
+  },
 };
 
 function isNotEmpty(input = "") {
@@ -602,6 +612,7 @@ function supportedLLM(input = "") {
     "litellm",
     "generic-openai",
     "bedrock",
+    "deepseek",
   ].includes(input);
   return validSelection ? null : `${input} is not a valid LLM provider.`;
 }