Skip to content
Agentmart
Submit a tool

supervise-agent-pipeline

Supervise a durable multi-phase agent worker, classify execution state, handle bounded approvals and provider fallback, and advance only after validating…

Skill by CraigThomasParsons

IndexedMIT★ 1
Write a review

Chalk Labs

Need help deploying supervise-agent-pipeline?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/craigthomasparsons-supervise-agent-pipeline

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Supervise a durable multi-phase agent worker, classify execution state, handle bounded approvals and provider fallback, and advance only after validating structured evidence. Use for tmux workers, phase pipelines, long autonomous runs, stalled workers, or resumable evidence-gated agent handoffs.

Use cases: data, agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md and code. Checked Nov 1, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found
Runs as
Python package

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/supervise-agent-pipeline

What's inside

Skill name
supervise-agent-pipeline
Description
Supervise a durable multi-phase agent worker, classify execution state, handle bounded approvals and provider fallback, and advance only after validating structured evidence. Use for tmux workers, phase pipelines, long autonomous runs, stalled workers, or resumable evidence-gated agent handoffs.
License
MIT
Instructions
46 lines in SKILL.md · view on GitHub

Before installing

  • Ships executable scripts

Files (4)

  • SKILL.md
  • openai.yaml · agents/
  • execution-states.md · references/
  • classify_execution.py · scripts/

SKILL.md (first lines)

# Supervise Agent Pipeline

Supervise the worker; never become the worker.

## Setup

Gather the packet, capability grant, worker adapter, ordered phases, evidence contract, polling cadence, and durable state location. Every phase must have a stable ID and resume boundary.

## Tick

Perform exactly one cycle:

1. Capture an adapter-neutral observation: process liveness, activity, approval request, provider error, gate result, receipt reference, and idle streak.
2. Classify it with `scripts/classify_execution.py` or an equivalent deterministic adapter.
3. Take at most one action:
   - `WORKING`: wait.
   - `WAITING_APPROVAL`: approve only when the action is explicitly in the grant; otherwise escalate.
   - `WAITING_PROVIDER`: hand off once to an allowed fallback with identical packet, phase, grant, and evidence requirements.
   - `BLOCKED_GATE`, `BLOCKED_SCOPE`, `FAILED_INFRASTRUCTURE`: stop and report evidence.
   - `IDLE`: wait for a second observation.
   - `STALLED`: inspect process health, then nudge once or restart from the phase boundary.
   - `DONE`: validate the structured receipt before advancing.
4. Persist the observation, classification, action, and idempotency key.

Run the tick from one durable recurring mechanism. Stop it after terminal completion or escalation.

## Phase transition

- Start each phase in a fresh context.
- Repeat repository, packet, phase, branch, base revision, write surface, prohibited actions, evidence commands, and receipt destination.
- Preserve user-authored input in interactive adapters.
- Require the worker to write a receipt, then optionally print a short adapter sentinel containing its path and digest.
- Do not trust a sentinel until the receipt validates.
- Return to `$coordinate-cross-project-work` when the packet phase completes.

## Tmux adapter lessons

- Type and submit in separate operations so prompts can be verified before execution.
- Treat subagent-wait markers as working, not idle.
- Require two idle observations before declaring a stall.
- Detect a dead worker process before nudging.
- Keep tmux scrollback observational; durable protocol state belongs outside the pane.

Read [references/execution-states.md](references/execution-states.md) for classification precedence.

From CraigThomasParsons/TheGiteaCodeFactory, MIT.

More from CraigThomasParsons/TheGiteaCodeFactory

Signals

Stars
1
Forks
0
Open issues
3
Last commit (this folder)
Oct 2, 2026 (1 month ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from CraigThomasParsons

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get supervise-agent-pipeline running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying supervise-agent-pipeline?

CallGet help
Cite this page

supervise-agent-pipeline skill. Agent Mart. https://agentmart.dev/skills/craigthomasparsons-supervise-agent-pipeline (last updated 2026-10-24)