Taskmaster

Taskmaster

Taskmaster is a completion guard for coding agents.

It addresses a common failure mode: the agent makes partial progress, writes a
summary, and stops before the user goal is actually finished.

Philosophy

Taskmaster is built around one idea: progress is not completion.

- Evidence over narrative:
  The agent should not be allowed to stop based on a convincing summary alone.
  Completion must be explicit and machine-checkable.
- Same-session recovery:
  When a turn is incomplete, the right move is to continue in the same running
  session, not restart from scratch.
- Goal re-anchoring:
  Compliance prompts force the model back to the user’s actual request, not its
  own local notion of “good enough”.
- Automation-safe signaling:
  A deterministic done token makes completion parseable for wrappers and
  CI-style flows.

Core Contract

A run is complete only when the assistant emits:

If that token is missing at stop time, Taskmaster blocks stop and pushes the
current turn to continue. Codex monitoring stays active for later turns in the
same long-lived session.

Enforcement Prompt

Taskmaster uses one shared compliance prompt for both Codex and Claude.

- Codex: the wrapper/injector path injects this shared prompt back into the
  same running session when stop conditions are not met.
- Claude: the Stop hook returns this same shared prompt as the block reason.

The shared prompt source lives in taskmaster-compliance-prompt.sh.

How It Works

- Codex path:
  - Runs through a wrapper (codex shim / codex-taskmaster launcher).
  - Enables Codex session logs.
  - Watches task_complete / turn_complete events.
  - If done token is missing, injects a continuation prompt into the same
    running Codex process via expect PTY.
  - A done token suppresses injection for…