It gives you six free, offline, read-only MCP tools for explaining, drafting, and checking Quiet File Team setup guidance, with no network, file access, execution, scheduling, or messaging.
explain_method: Explain the quiet-bus method, trust boundaries, idle behavior, and durable tasks by topic.
generate_team_config: Draft reviewed team.conf defaults from a roster and IANA time zone, with lead as coordinator and warnings.
generate_early_exit: Draft a quoted early-exit shell block for a specified role, roster, and team directory.
validate_layout_manifest: Compare caller-supplied relative paths against required free-core paths, reporting missing paths and limitations without inspecting the filesystem.
draft_workflow: Draft a fictional lead/builder/helper practice workflow with steps, acceptance checks, and warnings.
draft_owner_batch: Deduplicate explicit owner questions into a bounded review batch and preserve unresolved task references.
All outputs are structured JSON drafts; they do not prove runtime safety, authorization, installation, or real filesystem state.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Quiet File Team MCP draft a team config with a lead-only coordinator and default zones"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Give Claude, Cursor or any MCP client six free offline tools that explain, configure and check a Quiet File Team (several bots sharing one computer through files). Nothing leaves your machine: no network calls, no file reads, no paid APIs.
This project is a module of the Panda Federation — public-data tools, MCP servers, and local-first apps in one catalog. See the rest of the work, including books when they publish and the matching free or paid editions of this tool, at the official site.
Follow on X: @PandaFedHQ
Related MCP server: Enlace
uvx quiet-file-team-mcpInstall uv (Python 3.11+); no token or account is needed.
Add the server to your MCP client with the JSON below (Cursor, Claude Desktop or any stdio client).
Ask, e.g. "Explain the quiet file team method, then draft a config for lead, builder and helper in Europe/London." Use the output with the free Bash core.
Field | Value |
Purpose | Offline guidance for the Quiet File Team method: explain it, draft config and early-exit text, validate a supplied layout, draft a practice workflow, batch owner questions. |
Use when | Setting up or reviewing a file-first multi-bot team on one shared Linux machine. |
Do not use when | You need the scripts themselves (use the Bash core), real filesystem inspection, scheduling or messaging. |
Install |
|
Tools |
|
Outputs | Structured JSON per tool (output schemas included). |
Side effects and cost | None: read-only, non-destructive, idempotent, closed-world. Free. |
uvx quiet-file-team-mcpFrom this repo instead of PyPI:
uvx --from git+https://github.com/pandadad0214/quiet-file-team-mcp.git quiet-file-team-mcpA GitHub Release also provides the wheel, source archive and checksums: https://github.com/pandadad0214/quiet-file-team-mcp/releases/latest
For Cursor, Claude Desktop or another stdio MCP client:
{"mcpServers":{"quiet-file-team":{"command":"uvx","args":["quiet-file-team-mcp"]}}}Official MCP Registry name for this stdio package: io.github.pandadad0214/quiet-file-team-mcp (server.json). Registry clients install the PyPI package with uvx quiet-file-team-mcp. The git command above stays the install path from this repo.
The server requires Python 3.11+ and uv on the client. It computes text from bounded supplied inputs. It does not read arbitrary files, call network APIs, execute generated shell, install software or schedule anything. No paid APIs or Actor wrapper; latency is unbenchmarked. Tool outputs are drafts, not authorization or proof of runtime safety.
Tool | Use |
explain_method | Concepts and trust limits; choose before an exercise |
generate_team_config | Review safe roster/zone defaults; lead-only coordinator |
generate_early_exit | Draft a quoted block for a supplied role/path; outer routine must stop |
validate_layout_manifest | Compare supplied relative paths; no actual filesystem inspection |
draft_workflow | Fictional default-roster practice steps, not operational history |
draft_owner_batch | Deduplicate explicit owner questions and retain unresolved task references |
All tools are read-only, non-destructive, idempotent and closed-world. Structured output schemas accompany each. Optional task summaries are not evidence of a decision; no tool sends a message. Input privacy guards are heuristic, not comprehensive secret detection.
Edition | Price | Where |
Free Bash core (the scripts) | Free | |
This MCP companion (guidance only) | Free | PyPI and this repo |
Agensi setup kit skill | Paid (live price on the listing) | |
Apify setup kit generator (configured ZIP) | Paid, per generated kit (live price on the Store) |
MCP → free core: curl -fsSL https://github.com/pandadad0214/quiet-file-team/archive/refs/tags/v0.1.0.tar.gz | tar xz && mv quiet-file-team-0.1.0 team, set TEAM_DIR, follow its README. This MCP does not install those scripts.
Free → Agensi skill: load the skill in your agent; it guides install, rule packs and acceptance checks in your own folder.
Free → Apify generator: fill Roster (TEAM_IDS), IANA time zone (TEAM_TZ) and Coordinator (lead), allow a max charge at least the per-kit price shown on the Store page, Start, then download OUTPUT.zip from the run's key-value store into a new folder.
Paid → free: keep data backed up, use a separate clean core and repoint routines deliberately; never merge fictional logs.
python -m venv .venv
.venv/bin/pip install -e . pytest build
.venv/bin/pytest
.venv/bin/python -m buildCode is under src/quiet_file_team_mcp/server.py. Six deterministic tools use the MCP Python SDK with validated schemas and all four annotations. Runtime code has no subprocess or network client and no file-opening functions. Layout checks consume only caller-supplied path lists. MIT licensed.
Structure supplied owner-decision questions into a bounded review batch; unlike draft_workflow, organize decisions rather than simulate job execution. Use before an authorized owner review, never to send messages, infer approval or invent answers. Free offline text processing, no network or paid API and unbenchmarked short latency. Identical explicit questions merge task ids; missing questions or cap overflow preserve unresolved references. Returns questions, unresolved ids and warnings; summaries alone do not establish a decision.
Inputs: items (required), max_questions. Returns: warnings, questions, unresolved_task_ids
Draft a fictional lead/builder/helper task, board and pair exercise; unlike explain_method, return practice steps and acceptance checks. Use only in an empty disposable default-roster folder, not as production history or an executable fixture. Free offline template computation with no network or paid API, bounded work and unbenchmarked latency. Objective is treated as quoted prose, never executed; sensitive-shaped text is rejected heuristically. Returns scenario, ordered steps, acceptance checks and warnings.
Inputs: objective. Returns: steps, scenario, warnings, acceptance
Explain the file-first quiet-bus method and trust limits; unlike draft_workflow, return concepts rather than a practice sequence. Use before adoption or to clarify idle/durable behavior, not to install software or create schedules. Free offline text computation; no paid APIs or network calls, small bounded work with unbenchmarked latency. Topic selects emphasis, not an integration. Returns principles, prerequisites, limits and the free-core link.
Inputs: topic. Returns: limits, free_core, principles, prerequisites
Draft a quiet early-exit shell block for an existing routine; unlike generate_team_config, leave roster settings unchanged. Use while manually wiring a routine, not to schedule it, inspect files or prove the enclosing agent stops. Free offline text computation, no paid API or network calls; short bounded work, latency unbenchmarked. Role must belong to roster; team_dir is quoted as literal text and is never read. Match exactly NO WORK, not exit status. Returns block and enclosing-routine warnings.
Inputs: role (required), roster (required), team_dir (required). Returns: block, warnings
Draft reviewed team.conf defaults for a fresh local core; unlike generate_early_exit, produce roster settings rather than routine instructions. Use before manual installation, not for secrets, arbitrary shell fragments or modifying existing logs. Free offline text computation with no paid API and unbenchmarked short latency. Coordinator must be lead and occur in roster; environment overrides still win, and target zone support must be checked locally. Returns config text, normalized roster and warnings; writes nothing.
Inputs: roster (required), zone, coordinator. Returns: zone, roster, warnings, config_text, +1 more
Compare a supplied file manifest with required free-core paths; unlike generate_team_config, report omissions rather than draft settings. Use to review a known path list, not to read a filesystem, validate contents or prove runtime safety. Free offline computation, no network or paid API; bounded list processing with unbenchmarked latency. Absolute paths, traversal and oversized lists fail rather than being silently normalized. Returns completeness, missing paths and limitations; extras do not prove installation.
Inputs: paths (required). Returns: missing, complete, limitations, present_required
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 6 tools
Each description explicitly contrasts itself with a sibling tool ("unlike generate_team_config...", "unlike explain_method..."), giving an agent a clear routing map. The only real friction is the generate_* vs draft_* families, which all produce text artifacts and could be momentarily confused before reading the contrast clause.
All names are snake_case verb_noun, which is consistent and readable. However, four different verbs (explain, generate, validate, draft) cover six tools, and generate/draft are used near-synonymously for 'produce text output,' a minor semantic inconsistency.
Six tools is well within the ideal 3-15 range and each covers a distinct artifact: explanation, config, exit block, manifest check, practice scenario, decision batch. Nothing feels padded or missing at the count level.
The surface covers the offline-text-computation lifecycle: explain, generate config, generate routine block, validate manifest, practice, and structure decisions. Gaps exist (no tool to verify zone/environment support or to validate a generated config before install), but they are consistent with the stated 'writes nothing, no filesystem reads' boundary.
Read-only AI coding tools for change verification, release readiness, capacity, and guidance.
7 anonymous read-only tools. Authority-bounded autonomous onboarding guide. No financial authority.
Read-only game, setup, place, evidence and travel decision tools with explicit provenance.
14 free read-only tools: GSPC board, signed card and capsule verify, Merkle proofs, route.
