L1 Improvement
- Introduced Skill Forge sectioning with explicit guardrails, validation, and completion gates for playbook design.
- Added prompt-architect constraint extraction, ceiling discipline, and structured IO contracts for repeatable playbooks.
STANDARD OPERATING PROCEDURE
Purpose
Create actionable playbooks with phased steps, triggers, roles, evidence requirements, and completion criteria that teams can execute consistently.
Trigger Conditions
- Positive: need a new playbook, refactoring an existing one, aligning multiple teams to a standard procedure.
- Negative/reroute: single prompt rewrites (prompt-architect) or agent creation (agent-creator/agent-creation).
Guardrails
- Define start/stop triggers and success metrics for each phase.
- Keep outputs English-only with explicit confidence ceilings.
- Include evidence checkpoints and rollback/abort conditions to avoid unsafe execution.
- Make playbooks tool-agnostic unless a tool requirement is explicit.
Execution Phases
- Intake & Constraints: Capture objective, scope, actors, tools, and risks; classify HARD/SOFT/INFERRED constraints.
- Structure: Lay out phases with entry/exit criteria, roles, and artifacts.
- Detailing: Add step-by-step actions, decision points, and communication cadences.
- Validation: Dry-run or table-top the playbook; record findings, gaps, and ceilings.
- Delivery: Provide the finalized playbook, change log, and adoption guidance.
Pattern Recognition
- Incident response → emphasize triage, containment, comms, and postmortem steps.
- Release management → focus on pre-flight checks, gates, rollback plans.
- Research/analysis → define evidence collection, synthesis, and decision checkpoints.
Advanced Techniques
- Embed checklists and timeboxes for high-stress scenarios.
- Provide branching paths for common forks (e.g., severity levels).
- Include "stop the line" rules with authority levels.
Common Anti-Patterns
- Ambiguous ownership or missing role definitions.
- No rollback/abort conditions.
- Lack of evidence checkpoints or confidence ceilings.
Practical Guidelines
- Keep steps concise; prefer numbered lists and clear handoffs.
- Provide templates for logs/updates to ensure consistent communication.
- Note prerequisites and environmental assumptions explicitly.
Cross-Skill Coordination
- Upstream: prompt-architect for clarity; cognitive-lensing for alternative playbook framings.
- Parallel: meta-tools for tool alignment; base-template-generator for any supporting scripts.
- Downstream: agent-selector/agent-creator when playbooks require agent execution; recursive-improvement for continuous tuning.
MCP Requirements
- Optional memory MCP to store runbooks and lessons learned; tag WHO=playbook-architect-{session}, WHY=skill-execution.
Input/Output Contracts
inputs:
objective: string # required goal
scope: string # required boundaries
actors: list[string] # required roles/owners
constraints: list[string] # optional constraints and tools
outputs:
playbook: file # phased steps with triggers and checkpoints
validation_notes: file # dry-run findings and gaps
adoption_plan: summary # rollout steps and responsibilities
Recursive Improvement
- After first execution, run recursive-improvement on feedback to refine phases and guardrails.
Examples
- Draft an incident response playbook for P1 outages with severity-based branches and comms templates.
- Refine a release playbook to add feature flag rollouts and rollback checkpoints.
Troubleshooting
- Steps unclear → rewrite with subject-verb-object and explicit owners.
- Playbook too long → modularize by phase and severity.
- Missing safety nets → add abort conditions and rollback steps.
Completion Verification
Confidence: 0.70 (ceiling: inference 0.70) - Playbook Architect SOP rewritten with Skill Forge cadence and prompt-architect ceilings.
1---2name: playbook-architect3description: Design, refactor, and validate operational playbooks with clear phases, triggers, and evidence-backed checkpoints.4---5
6
7
8
9### L1 Improvement
10- Introduced Skill Forge sectioning with explicit guardrails, validation, and completion gates for playbook design.
11- Added prompt-architect constraint extraction, ceiling discipline, and structured IO contracts for repeatable playbooks.
12
13## STANDARD OPERATING PROCEDURE
14
15### Purpose
16Create actionable playbooks with phased steps, triggers, roles, evidence requirements, and completion criteria that teams can execute consistently.
17
18### Trigger Conditions
19- Positive: need a new playbook, refactoring an existing one, aligning multiple teams to a standard procedure.
20- Negative/reroute: single prompt rewrites (prompt-architect) or agent creation (agent-creator/agent-creation).
21
22### Guardrails
23- Define start/stop triggers and success metrics for each phase.
24- Keep outputs English-only with explicit confidence ceilings.
25- Include evidence checkpoints and rollback/abort conditions to avoid unsafe execution.
26- Make playbooks tool-agnostic unless a tool requirement is explicit.
27
28### Execution Phases
291. **Intake & Constraints**: Capture objective, scope, actors, tools, and risks; classify HARD/SOFT/INFERRED constraints.
302. **Structure**: Lay out phases with entry/exit criteria, roles, and artifacts.
313. **Detailing**: Add step-by-step actions, decision points, and communication cadences.
324. **Validation**: Dry-run or table-top the playbook; record findings, gaps, and ceilings.
335. **Delivery**: Provide the finalized playbook, change log, and adoption guidance.
34
35### Pattern Recognition
36- Incident response → emphasize triage, containment, comms, and postmortem steps.
37- Release management → focus on pre-flight checks, gates, rollback plans.
38- Research/analysis → define evidence collection, synthesis, and decision checkpoints.
39
40### Advanced Techniques
41- Embed checklists and timeboxes for high-stress scenarios.
42- Provide branching paths for common forks (e.g., severity levels).
43- Include "stop the line" rules with authority levels.
44
45### Common Anti-Patterns
46- Ambiguous ownership or missing role definitions.
47- No rollback/abort conditions.
48- Lack of evidence checkpoints or confidence ceilings.
49
50### Practical Guidelines
51- Keep steps concise; prefer numbered lists and clear handoffs.
52- Provide templates for logs/updates to ensure consistent communication.
53- Note prerequisites and environmental assumptions explicitly.
54
55### Cross-Skill Coordination
56- Upstream: prompt-architect for clarity; cognitive-lensing for alternative playbook framings.
57- Parallel: meta-tools for tool alignment; base-template-generator for any supporting scripts.
58- Downstream: agent-selector/agent-creator when playbooks require agent execution; recursive-improvement for continuous tuning.
59
60### MCP Requirements
61- Optional memory MCP to store runbooks and lessons learned; tag WHO=playbook-architect-{session}, WHY=skill-execution.
62
63### Input/Output Contracts
64```yaml
65inputs:
66 objective: string # required goal
67 scope: string # required boundaries
68 actors: list[string] # required roles/owners
69 constraints: list[string] # optional constraints and tools
70outputs:
71 playbook: file # phased steps with triggers and checkpoints
72 validation_notes: file # dry-run findings and gaps
73 adoption_plan: summary # rollout steps and responsibilities
74```
75
76### Recursive Improvement
77- After first execution, run recursive-improvement on feedback to refine phases and guardrails.
78
79### Examples
80- Draft an incident response playbook for P1 outages with severity-based branches and comms templates.
81- Refine a release playbook to add feature flag rollouts and rollback checkpoints.
82
83### Troubleshooting
84- Steps unclear → rewrite with subject-verb-object and explicit owners.
85- Playbook too long → modularize by phase and severity.
86- Missing safety nets → add abort conditions and rollback steps.
87
88### Completion Verification
89- [ ] Phases include entry/exit criteria, roles, and success metrics.
90- [ ] Evidence checkpoints and rollback/abort conditions documented.
91- [ ] Confidence ceiling stated; validation notes captured.
92- [ ] Adoption guidance provided (who uses it, when, and how).
93
94Confidence: 0.70 (ceiling: inference 0.70) - Playbook Architect SOP rewritten with Skill Forge cadence and prompt-architect ceilings.
majiayu000