Skip to content

checklist-templates.md

Latest commit

 

History

History
121 lines (101 loc) · 7.37 KB

checklist-templates.md

File metadata and controls

121 lines (101 loc) · 7.37 KB

Checklist Templates — Skill Instruction File

Version: 1.0.0 | Documents the dynamic checklist system in SKILL.md

Read when adapting the checklist format for a new platform or use case.

Canonical platform behavior: docs/platform-capabilities.md
Execution modes (CoWork): docs/EXECUTION_MODES.md
Mandatory exclusions (scope law): docs/MANDATORY_EXCLUSIONS.md

Purpose

The dynamic session checklist is the live progress board that the system prints at the start of every phase and updates after each scoring gate. It serves three functions: it keeps the user informed of session state, it prevents the system from skipping phases silently, and it provides placeholder structure that constrains the LLM from hallucinating values into unfilled fields.

Checklist Design Principles

The checklist must always reflect the actual state of the session, not an aspirational or projected state. It is updated after each scoring gate, not before. A phase is never marked Complete until the user has confirmed a score of 5 for that phase.

Placeholder tokens in the checklist header ([CANDIDATE_NAME], [ROLE_TITLE], [COMPANY_NAME]) are populated at session start from the applicant profile. If a value is unknown at session start, the token remains visible — this is intentional. It signals to both the user and the system that a required value has not yet been confirmed.

Full Checklist Template

╔════════════════════════════════════════════════════════════════════╗
║          JOB APPLICATION ENGINE — SESSION STATUS BOARD            ║
╠════════════════════════════════════════════════════════════════════╣
║  Phase 0 │ Job Discovery          │ [STATUS] │ Score: [ /5]       ║
║  Phase 1 │ Company Intelligence   │ [STATUS] │ Score: [ /5]       ║
║  Phase 2 │ Fit Analysis           │ [STATUS] │ Score: [ /5]       ║
║  Phase 3 │ Clarifying Intake      │ [STATUS] │ Score: [ /5]       ║
║  Phase 4 │ Application Package    │ [STATUS] │ Score: [ /5]       ║
║  Phase 5 │ Writing Quality Pass   │ [STATUS] │ Score: [ /5]       ║
║  Phase 6 │ Governance Gate        │ [STATUS] │ Score: [ /5]       ║
║  Phase 7 │ Post-Submission Loop   │ [STATUS] │ Score: [ /5]       ║
╠════════════════════════════════════════════════════════════════════╣
║  Active Phase : [PHASE_NAME]                                      ║
║  Current Gate : [GATE_DESCRIPTION]                                ║
║  Awaiting     : [WHAT_IS_NEEDED_FROM_USER]                        ║
║  Candidate    : [CANDIDATE_NAME]                                  ║
║  Session Role : [ROLE_TITLE]     Company: [COMPANY_NAME]          ║
╚════════════════════════════════════════════════════════════════════╝

Status codes:
  ⏳ Pending   — phase has not started
  🔄 Active    — phase is currently running
  🔁 Revision  — phase output scored below 5, revision in progress
  ✅ Complete  — phase scored 5, gate passed, advancing
  🚫 Blocked   — phase cannot start until a preceding gate clears

Example — Mid-Session State (Alex M., fictional)

╔════════════════════════════════════════════════════════════════════╗
║          JOB APPLICATION ENGINE — SESSION STATUS BOARD            ║
╠════════════════════════════════════════════════════════════════════╣
║  Phase 0 │ Job Discovery          │ ✅ Complete │ Score: 5/5      ║
║  Phase 1 │ Company Intelligence   │ ✅ Complete │ Score: 5/5      ║
║  Phase 2 │ Fit Analysis           │ ✅ Complete │ Score: 5/5      ║
║  Phase 3 │ Clarifying Intake      │ 🔄 Active   │ Score: [ /5]   ║
║  Phase 4 │ Application Package    │ ⏳ Pending  │ Score: [ /5]   ║
║  Phase 5 │ Writing Quality Pass   │ ⏳ Pending  │ Score: [ /5]   ║
║  Phase 6 │ Governance Gate        │ ⏳ Pending  │ Score: [ /5]   ║
║  Phase 7 │ Post-Submission Loop   │ ⏳ Pending  │ Score: [ /5]   ║
╠════════════════════════════════════════════════════════════════════╣
║  Active Phase : Clarifying Intake                                 ║
║  Current Gate : Product trial observation (awaiting user input)   ║
║  Awaiting     : User to complete 10-min product trial and return  ║
║  Candidate    : Alex M.                                           ║
║  Session Role : Senior Product Manager   Company: CloudBase Inc   ║
╚════════════════════════════════════════════════════════════════════╝

Scoring Gate Template

Print this exact format after every phase output. Do not paraphrase it.

── PHASE [N] REVIEW GATE ──────────────────────────────────────────
Please respond with:
  1. What was correct
  2. Your score from 1 to 5  (5 required to proceed)
  3. What you expected for a perfect result at this phase
If score < 5, state what to revise. This phase reruns before advancing.
───────────────────────────────────────────────────────────────────

Revision Loop Behaviour

When a score below 5 is received:

  1. Update the phase status to 🔁 Revision in the checklist.
  2. Read the user's stated expectation for a perfect result.
  3. Revise only the elements identified — do not rerun the entire phase.
  4. Present the revised output.
  5. Present the scoring gate again.
  6. If the revised score is 5, update status to ✅ Complete and advance.
  7. If the revised score is below 5 for a second time, ask the user whether to continue revising or to proceed despite the lower score. Do not force advancement. Do not stall indefinitely.

Platform Notes

Claude.ai: print the checklist as formatted text. It renders in monospace where code blocks are used, which preserves the box drawing characters.

Claude CoWork: generate the checklist as a static artifact that is updated at each gate. The artifact persists across the subagent coordination flow.

Manus: maintain the checklist as a text block in the session instruction context. Update it in place at each gate by replacing the previous version.