execute-detailed

Executes one bounded implementation plan with verification, acceptance checks, and PR publication.

2|1|Updated Jun 25, 2026
One-click install
npx skills add https://github.com/Consiliency/agent-harness --skill execute-detailed-consiliency
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: execute-detailed
Source: https://github.com/Consiliency/agent-harness/tree/main/phase-loop-skills/execute-detailed
Command: npx skills add https://github.com/Consiliency/agent-harness --skill execute-detailed-consiliency

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? It turns a detailed implementation plan into verified, published code changes without scope creep, unsafe git operations, or unverified closeouts, so a single agent thread can carry a bounded change end to end. ## Core Features & Use Cases - Bounded plan execution: Reads a detailed plan artifact and applies only its declared ## Changes section, stopping with a diagnostic if work falls outside the plan. - Verification and acceptance reduction: Runs every command in the plan's ## Verification section and checks each ## Acceptance criteria item, reporting pass/fail status per item. - Safe git workflow: Establishes a dedicated worktree and branch before editing, enforces a git safety invariant, and publishes a draft or ready PR via the runtime publication primitive. - Mandatory reflection closeout: Writes a structured reflection and updates the plan manifest lifecycle for every non-trivial run. - Use Case: After a planning skill produces a detailed plan for a bug fix, invoke this skill to implement the changes in an isolated worktree, run the test suite, and open a pull request with the verified diff. ## Quick Start Execute the detailed plan at the given artifact path, verify it against its acceptance criteria, and publish the result as a pull request.

Frequently Asked Questions about execute-detailed

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I execute a detailed implementation plan with an AI agent?▼

Provide the path to a detailed plan artifact containing Changes, Verification, and Acceptance criteria sections. The skill reads the plan, applies the declared changes in a safe worktree, runs the verification commands, and reports acceptance status before publishing a PR.

What happens when plan verification fails during execution?▼

The skill attempts one local diagnosis and repair, then reports the stable diagnostic verification failed if the command still fails or falls outside the plan's scope. The plan manifest lifecycle is marked failed and the closeout records the failure.

Does this skill commit directly to the main branch?▼

No. A git safety invariant forbids committing to main or protected branches, dirty checkouts, or branches the run does not own. It creates a dedicated worktree and branch off a fresh base ref and publishes a PR from there.

When does the skill open a draft versus a ready pull request?▼

It opens a draft PR when dependencies remain or verification was partial or skipped, and a ready PR only when verification is complete. If publication prechecks fail, it reports publication blocked and preserves the work on the safe branch.

What are the limitations of bounded plan execution?▼

The skill cannot implement anything outside the plan's declared Changes section; encountering unlisted files or behavior stops the run with a dirty worktree diagnostic. It also defers commit and closeout entirely when a runner or pipeline owns closeout.