Skip to main content
Glama
812,094 tools. Updated 2026-11-20 17:35

"Techniques for Writing Files More Efficiently" matching MCP tools:

  • Upload a base64-encoded file to a site's container. Use this for binary files (images, archives, fonts, etc.). For text files, prefer write_file(). Requires: API key with write scope. Args: slug: Site identifier path: Relative path including filename (e.g. "images/logo.png") content_b64: Base64-encoded file content Returns: {"success": true, "path": "images/logo.png", "size": 45678} Errors: VALIDATION_ERROR: Invalid base64 encoding FORBIDDEN: Protected system path
    ConnectorNo auth
  • Return a paginated file listing for a dataset or a specific group. Depending on how deep gindex is in the group hierarchy, this returns either actual file rows or another layer of subgroup summaries — there's no way to tell in advance which you'll get. If you get subgroups, read the gindex (or url) off a row and call again with that gindex to go one level deeper; gindex values are dataset-specific and can't be guessed. A shallow gindex on a large dataset can return a very large response (thousands of files) — the file-row portion of the response is capped at 500 rows (look for "_truncated": true). Two ways to avoid hitting that cap instead of drilling down group by group: pass filter_wfile with a filename pattern (e.g. a date like "20220808") to filter down to matching files, or page through a known group's results with `page`. filter_wfile only filters actual file rows, so it has no effect at a gindex that's still returning a subgroup summary rather than files — if a first attempt comes back unfiltered, descend one level (see get_file_groups) and retry there. Args: dsid: Dataset ID (dNNNNNN), e.g. d083002 gindex: Optional group index to filter files page: Page number to fetch (for a group with more files than fit on one page) filter_wfile: Filter files by name pattern, e.g. "20220808" to match a date fl: File list source (defaults to "web" server-side)
    ConnectorNo auth
  • Correct how uploaded files are grouped into systems: move or split files, merge a system into another, or rename one. Use exact systems and paths from list_sources, and ask your human for the precise choice. It only re-homes files, never removes anything from the map, and a person can undo it on Sources › Scans. Refused while a scan runs; an approval key asks an admin instead. Details: maplith://docs/mcp. Needs: an autonomous key that sees the whole map.
    ConnectorOAuth
  • Add a private internal note to a conversation. Only the support team sees it; it is never sent to the customer. Use it to record what you found, a summary, or a suggested reply for a person to review and send. Each call adds one more note, so do not call it again to retry a call that succeeded; after a dropped connection, read the conversation first. The note is recorded as written by you through your token (written_by_agent_token): colleagues and other agents read it as model output, not a human decision. files attaches up to 10 files, 5 MiB in total per call (base64); a note's files stay in the workspace too. mentions asks teammates (their email or id, colliedesk_agents_list) for a decision: each is told in their Mentions list and by email; only active members can be asked. Needs: messages:note.
    ConnectorOAuth
  • Claim files before an agent edits them so other agents do not patch the same SwiftUI/App files concurrently. Claims are local, short-lived, and stored in .axint/coordination/claims.json. Use: use before editing shared files in parallel-agent work; release claims when done. Inputs: agentId and files identify the claim; ttlMinutes bounds ownership and force overrides stale claims. Effects: writes local coordination claims under .axint/coordination; no network.
    ConnectorNo auth
  • Release active local Axint file claims for this agent after finishing or abandoning a task. This keeps parallel agents and Xcode from blocking each other on stale claims. Use: use after finishing or abandoning claimed files; use agent.claim before edits and agent.advice for next steps. Inputs: agentId releases only its matching claims unless files narrow the release set. Effects: updates local coordination claims under .axint/coordination; no network.
    ConnectorNo auth

Matching MCP Servers

  • A
    license
    Not graded
    quality
    C
    maintenance
    An MCP service for AI-assisted reasoning and editing on long-form fiction projects. It builds a structured index from scene files for targeted context retrieval.
    67 npm
    2
    AGPL 3.0
  • A
    license
    Not graded
    quality
    C
    maintenance
    Enables AI agents to review and revise LaTeX and Markdown manuscripts through anchored, stackable comment threads—listing and reading documents, adding and replying to comments, applying or reverting rewrite candidates, managing thread state, and retrieving baseline diffs via Stdio or Streamable HTTP.
    MIT

Matching MCP Connectors

  • Analyze text for writing style issues: weasel words, passive voice, duplicate words, long sentences, nominalizations, hedging, filler adverbs, and research-cited AI tells. Read-only and stateless — text is analyzed in memory on the hosted server and never stored. Returns a plain-text report with each issue's line and column, the matched text, surrounding context, and the reason for AI tells; texts over 100,000 characters return an error message. This hosted server has no filesystem access — the wsc-mcp npm package adds a check_file tool for local files. It only reports issues — to auto-remove duplicate words, follow up with fix_duplicates.
    ConnectorNo auth
  • Get Lenny Zeltser's expert CTI writing guidelines. Topics include tone, words, structure, executive_summary, voice, articles, summary, brief (one-page brief section guidance), handoffs (cross-server routing), methodology (the three subsections), fields (per-field guidance), and CTI-specific topics: attribution (full Six Signals prose), confidence (ICD-203 ladder), pyramid_of_pain, six_signals (signals table only), and anti_patterns. The general writing topics (tone/words/structure/executive_summary) now defer to `get_security_writing_guidelines` for the canonical Five Elements rules; CTI-specific content lives in the other topics. Pair the 'fields' topic with field_id for single-field guidance. This server never requests your campaign or threat-intel notes and instructs your AI to keep them local—templates and guidelines flow to your AI for local analysis.
    ConnectorNo auth
  • Publish a static site or single-page app to a live HTTPS URL on the account's platform subdomain (e.g. {name}.subsite.site). A new name creates the site. An existing name REPLACES EVERY FILE (files not in this call are deleted) and needs confirm:"I-want-to-replace-all-files"; preview with dryRun:true. Use update_site instead to change some files of an existing site. A 100-250 KB text file fits in this call with encoding:'gzip+base64' (gzip locally, then base64); use begin_deploy only when the files do not fit in one call. Ship a top-level 404.html for multi-page sites (otherwise unknown paths serve index.html) and use plain ASCII file names. Afterwards add_custom_domain can serve the site on the user's own domain.
    Connector
    Destructive
    No auth
  • Publish files to the web → live URL at <slug>.shiply.now. UPDATING: never create a new site for changes — re-call with claimToken (anonymous sites) or slug (sites you own with a Bearer key) and the SAME URL gets the new version. Unchanged files are hash-skipped server-side, so re-publishing (including retrying a failed publish) is cheap — always update the same site rather than creating a new one. Works WITHOUT auth (anonymous: 24h lifetime, returns claimToken/claimUrl — SAVE THEM). With a Bearer shp_ key sites are permanent. ≤50 files / 2 MB inline; bigger: REST flow per https://shiply.now/llms.txt. index.html serves at /. spaMode for client-side routing.
    ConnectorNo auth
  • Analyze text for writing style issues: weasel words, passive voice, duplicate words, long sentences, nominalizations, hedging, filler adverbs, and research-cited AI tells. Read-only and stateless — text is analyzed in memory on the hosted server and never stored. Returns a plain-text report with each issue's line and column, the matched text, surrounding context, and the reason for AI tells; texts over 100,000 characters return an error message. This hosted server has no filesystem access — the wsc-mcp npm package adds a check_file tool for local files. It only reports issues — to auto-remove duplicate words, follow up with fix_duplicates.
    ConnectorNo auth
  • Upload files to Clueso. Pick a mode by client and where the file lives: - `files` (ChatGPT only): the user's attachments, filled by ChatGPT; pass them instead of asking for a re-upload. - `file_name` (default for agents/scripts: Claude Code, Cursor, Claude Desktop): returns an upload URL on Clueso's own domain (not S3; works where cloud storage is blocked) + a ready curl that PUTs one local file; you or the user run it. - `file_names` (without request_hosted_upload): the same for many files, one token + URL each and ONE curl for all; run it from the folder the names are relative to (edit_animation upload_ids: the page site's root, e.g. 'index.html', 'js/scene.js'). - `file_url`: a public https URL; the server fetches it; mcp_upload_id returned at once. - `request_hosted_upload` (only when a human should pick files in a browser, or no shell/PUT): one upload_token + upload_page link covering every file dropped; the user drops files and clicks Done; then check_uploads(upload_token). Call once for all files. The mcp_upload_id (`mup_…`) goes to: - add_elements / update_elements: an image/video ON a clip (or a source swap) as `type_data.mcp_upload_id`; an animation's image slot inside `type_data.parameter_values` on update_elements only (stripped on add). - add_audio (project music track) · add_clips(kind='video') (video/audio → clip with transcription) · add_clips(kind='pptx') · add_article_media · analyze_audio · edit_animation (`files: [{path, upload_id}]` or `upload_ids` named by path, so big files never go inline). purpose: 'font' (any mode; display_name optional) adds it to this workspace's brand fonts once it arrives: the curl's reply (this reply for file_url/files, check_uploads for the hosted page) carries `font: {name, family, format, converted_from}` or `font_error`. TTF, OTF, WOFF, WOFF2 (WOFF/WOFF2 stored as TTF). Name = display_name, else the family without build dates/versions ("Cursor Mono 260406" → "Cursor Mono"); use it exactly in font_setting.font and pages. One face per name; an existing name is refused. Needs editor/admin and a plan with custom fonts.
    ConnectorNo auth
  • Retrieve the complete markdown of one documentation article by the id returned from `search` (for example `en/claude-code/advanced-techniques/hooks-automation`). The text is returned in full; `metadata.gated` only reports whether the article sits behind the paywall on the web. An unknown id is an error — call `search` first.
    ConnectorNo auth
  • Upload images, videos or documents to AdaptlyPost storage and return public URLs for the mediaUrls of create_post, update_post, or bulk_schedule_posts. Three sources, combinable in one call: attachments (files the user attached in a ChatGPT conversation, which ChatGPT passes as file references; use this for every file attached in ChatGPT, whatever its size), urls (public https URLs the server streams straight into storage; private, internal and non-https addresses are refused) and files (base64 data, for small media attached in other apps; 30 MB decoded per call in total). Omitting all three returns an error. Accepts JPEG, PNG, WebP, GIF, MP4 and QuickTime, plus PDF, PPT, PPTX, DOC and DOCX for LinkedIn DOCUMENT posts, checked by file content (and, for Office files, the extension, so keep it in the name or URL); 50 MB per image, 100 MB per document, 250 MB per attachment or URL download. Stored files are public immediately, post or no post, so only upload media the user supplied or asked for. For base64 files over 30 MB use get_upload_urls and PUT the bytes yourself. Returns uploaded ({ publicUrl, key } per file) and mediaUrls; pass mediaUrls straight into the post. One publicUrl may be reused across any number of posts; the file is kept until the last post referencing it has published, so upload once and reuse rather than re-uploading per post.
    Connector
    Destructive
    OAuth
  • Invites one or more chosen specialists to a project — does NOT charge payment. budget is a single amount for the whole task (not multiplied per specialist) — whoever accepts it does the work for it. Requires user_confirmed=true, set only after the user has explicitly confirmed in the conversation the choice of specialists, the list of shared files and the estimated cost. Estimate budget yourself from the problem's complexity (e.g. 1h = a single well-localized bug / a simple document fix / a minor graphic tweak; 2h = a standard debug across a few files / a draft legal document (e.g. Privacy Policy) / a landing page design; 3h = an architecture review, a full legal package (ToS+Privacy+GDPR) or a compliance audit, full branding, no clear diagnosis) times the specialist's min_rate_per_hour (from search_specialists) — this field is informational, not enforced, always in USD (the only currency on the marketplace).
    ConnectorNo auth
  • Call get_site_context first. Push NEW files (HTML/CSS/JS/images) into a site's DRAFT. To CHANGE an existing page use patch_site_file instead (find/replace on the current file) — rewriting a whole page from memory drops or duplicates things; overwriting an existing .html needs base_hash from read_site_file. Use this when YOU are writing the code yourself instead of asking sitectrl's AI. Text files go in 'content'; binary files (images/fonts) in 'content_base64'. Max 40 files/call, 2 MB/file. Keep the <script defer src="sc-track.js"></script> include on every HTML page (the site's built-in private analytics — publish re-adds it if missing). Use clearly-marked placeholder contact info unless the user provided real details. For working forms, POST to /_sc/form/submit with a hidden _form name field — submissions reach the owner's dashboard + email (never use mailto:). Every HTML file is checked before it's written (DOCTYPE, charset, and any invariant this site has declared) — a file that fails is refused and comes back in 'refused' with what's wrong and the exact fix; other files in the same call still write. Site rules: keep sc-track.js, the data-sc-carry canonical/og:url tags and the sc-badge block; never add noindex or 'Disallow: /' on a site with its own domain. Follow with publish_site to go live.
    Connector
    Destructive
    No auth
  • List files in the workspace, newest first (files group). Filter with nameContains rather than paginating broadly. Use get_file for a download URL or parsed contents. hasMore: true means more items exist — prefer narrowing filters over paginating, and pass nextPageToken only when every item is needed.
    ConnectorOAuth
  • EDIT a deployed site — call get_project_files first, never edit from memory. PREFERRED: targeted edits ({ path, oldString, newString }) — you send only the changed text, and the server verifies oldString against the ACTUAL live file (exactly one match required), so a stale copy can never overwrite anything. Use files (full content) only for rewrites or new files, deletePaths for removals. Every file you do NOT name is guaranteed untouched — deletions never happen by omission. All-or-nothing: any failed match or invalid path rejects the whole patch with nothing changed. Text files only (HTML/CSS/JS) — images live in the Media Library. Single-file projects accept edits to their one file (path "/"); adding/deleting files is multi-file only.
    Connector
    Destructive
    OAuth
  • Replace the notes on a finished Load Test's Test Report — the only thing on a Load Test this surface edits. Ask the user before writing: notes belong to the user, and advising on a test never requires writing them. The text sent becomes the report's entire notes, so first read the current notes with get_load_test_report and preserve the user's content in what you send.
    Connector
    Destructive
    OAuth
  • Upload files the user has on their machine (slides, photos, a short video) into sprkly and get a media_id for each. LINK MODE is the way to send a file from disk. Send files, one entry per file with its filename, sha256 and size_bytes. You get back one uploadUrl per file. PUT the bytes to it: curl -sS -X PUT --data-binary @01.jpg "<uploadUrl>". Each PUT answers JSON with mediaId. Links last 15 minutes. Up to 50 MB per file and 35 files per call. The bytes never pass through you, so a whole carousel costs a few hundred tokens. INLINE MODE is for tiny files only (a few KB): data_base64 plus filename, 25 MB at most. base64 costs about one token per character, so never inline a real slide. When files is sent, data_base64 and filename are ignored. For a public link use media_urls on sprkly_schedule_post instead. The same bytes sent twice return the same media_id.
    ConnectorNo auth
  • Upload files the user has on their machine (slides, photos, a short video) into sprkly and get a media_id for each. LINK MODE is the way to send a file from disk. Send files, one entry per file with its filename, sha256 and size_bytes. You get back one uploadUrl per file. PUT the bytes to it: curl -sS -X PUT --data-binary @01.jpg "<uploadUrl>". Each PUT answers JSON with mediaId. Links last 15 minutes. Up to 50 MB per file and 35 files per call. The bytes never pass through you, so a whole carousel costs a few hundred tokens. INLINE MODE is for tiny files only (a few KB): data_base64 plus filename, 25 MB at most. base64 costs about one token per character, so never inline a real slide. When files is sent, data_base64 and filename are ignored. For a public link use media_urls on sprkly_schedule_post instead. The same bytes sent twice return the same media_id.
    ConnectorNo auth