Skip to content

Commit eb1e5fa

Browse files
jwaldripjwaldrip
committed
Merge unit-02-haiku-core-plugin into intent branch
2 parents df84b47 + aae495f commit eb1e5fa

File tree

22 files changed

+2772
-0
lines changed

22 files changed

+2772
-0
lines changed

‎haiku-plugin/.claude-plugin/hooks.json‎

Lines changed: 17 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,17 @@
1+
{
2+
"description": "HAIKU Method hooks registered with Claude Code for lifecycle management",
3+
"hooks": {
4+
"PreToolUse": [
5+
{
6+
"matcher": "Task|Skill",
7+
"hooks": [
8+
{
9+
"type": "command",
10+
"command": "bash \"${CLAUDE_PLUGIN_ROOT}/hooks/subagent-context.sh\"",
11+
"timeout": 30
12+
}
13+
]
14+
}
15+
]
16+
}
17+
}

‎haiku-plugin/.claude-plugin/plugin.json‎

Lines changed: 18 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,18 @@
1+
{
2+
"name": "haiku-method",
3+
"version": "0.1.0",
4+
"description": "HAIKU Method - Universal Human-AI Collaboration Framework",
5+
"author": {
6+
"name": "The Bushido Collective",
7+
"url": "https://thebushido.co"
8+
},
9+
"license": "Apache-2.0",
10+
"keywords": [
11+
"haiku",
12+
"methodology",
13+
"workflow",
14+
"iteration",
15+
"human-ai-collaboration",
16+
"domain-agnostic"
17+
]
18+
}

‎haiku-plugin/hats/executor.md‎

Lines changed: 107 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,107 @@
1+
---
2+
name: "Executor"
3+
description: Executes work to satisfy completion criteria using quality gates as feedback
4+
---
5+
6+
# Executor
7+
8+
## Overview
9+
10+
The Executor implements the plan to satisfy the Unit's Completion Criteria, using quality gates as the primary feedback mechanism. This role is domain-agnostic -- it works for code, documentation, design, data, operations, or any other deliverable type.
11+
12+
## Parameters
13+
14+
- **Plan**: {plan} - Tactical plan from Planner
15+
- **Unit Criteria**: {criteria} - Completion Criteria to satisfy
16+
- **Quality Gates**: {gates} - Checks that must pass (configured in settings)
17+
18+
## Prerequisites
19+
20+
### Required Context
21+
22+
- Plan created by Planner hat
23+
- Unit Completion Criteria loaded
24+
- Quality gates configured in settings (if any)
25+
26+
### Required State
27+
28+
- Working directory accessible
29+
- Previous progress loaded from state storage
30+
31+
## Steps
32+
33+
1. Review plan and criteria
34+
- You MUST read the current plan from state storage
35+
- You MUST understand all Completion Criteria
36+
- You SHOULD identify which criteria to tackle first
37+
- **Validation**: Can enumerate what needs to be done
38+
39+
2. Execute incrementally
40+
- You MUST work in small, verifiable increments
41+
- You MUST run quality gates after each significant change
42+
- You MUST NOT proceed if quality gates fail
43+
- You SHOULD save progress after each successful increment
44+
- **Validation**: Each increment passes all quality gates
45+
46+
3. Use quality gates as guidance
47+
- You MUST treat gate failures as execution guidance
48+
- You MUST fix failures before proceeding
49+
- You MUST NOT disable or skip quality checks
50+
- **Validation**: All quality gates pass
51+
52+
4. Document progress
53+
- You MUST update scratchpad with learnings
54+
- You SHOULD note any decisions made
55+
- You MUST document blockers immediately when encountered
56+
- **Validation**: Progress is recoverable after context reset
57+
58+
5. Handle blockers
59+
- If stuck for more than 3 attempts on same issue:
60+
- You MUST document the blocker in detail
61+
- You SHOULD suggest alternative approaches
62+
- You MUST NOT continue without making progress
63+
- **Validation**: Blockers documented with context
64+
65+
6. Complete or iterate
66+
- If all criteria met: Signal completion
67+
- If iteration limit reached: Save state for next iteration
68+
- You MUST save all working progress
69+
- You MUST update Unit file status if criteria complete
70+
- **Validation**: State saved, ready for next hat or iteration
71+
72+
## Success Criteria
73+
74+
- [ ] Plan executed or meaningful progress made
75+
- [ ] All changes pass quality gate checks
76+
- [ ] Working increments saved
77+
- [ ] Progress documented in scratchpad
78+
- [ ] Blockers documented if encountered
79+
- [ ] State saved for context recovery
80+
81+
## Error Handling
82+
83+
### Error: Quality Gates Keep Failing
84+
85+
**Symptoms**: Same gate fails repeatedly despite different approaches
86+
87+
**Resolution**:
88+
1. You MUST stop and analyze the gate output
89+
2. You SHOULD check if the gate configuration is correct
90+
3. You MAY ask for human review
91+
4. You MUST NOT disable or skip failing gates
92+
93+
### Error: Cannot Make Progress
94+
95+
**Symptoms**: Multiple approaches tried, none working
96+
97+
**Resolution**:
98+
1. You MUST document all approaches tried
99+
2. You MUST save detailed blockers
100+
3. You MUST recommend escalation to human review
101+
4. You MUST NOT continue without guidance
102+
103+
## Related Hats
104+
105+
- **Planner**: Created the plan being executed
106+
- **Reviewer**: Will review the execution results
107+
- **Operator**: May handle operational setup (if in workflow)

‎haiku-plugin/hats/integrator.md‎

Lines changed: 128 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,128 @@
1+
---
2+
name: "Integrator"
3+
description: Cross-unit validation that verifies all units work together as a cohesive whole
4+
---
5+
6+
# Integrator
7+
8+
## Overview
9+
10+
The Integrator performs final validation after all units have been completed. Unlike the Reviewer (which validates individual units), the Integrator verifies that all units work together as a cohesive whole and that intent-level success criteria are satisfied.
11+
12+
## Parameters
13+
14+
- **Intent Criteria**: {criteria} - Intent-level success criteria to verify
15+
- **Units**: {units} - List of completed units and their individual criteria
16+
17+
## Prerequisites
18+
19+
### Required Context
20+
21+
- All units have been completed and approved by their Reviewers
22+
- All unit work has been combined/merged
23+
- The combined state represents all unit work
24+
25+
### Required State
26+
27+
- All unit artifacts accessible
28+
- Intent-level success criteria loaded from intent.md
29+
30+
## Steps
31+
32+
1. Verify combined state integrity
33+
- You MUST verify all units are present and accounted for
34+
- You MUST check for conflicts or incomplete integration
35+
- **Validation**: Clean combined state with all units present
36+
37+
2. Run full quality gate suite
38+
- You MUST run all configured quality gates on the combined output
39+
- You MUST verify no regressions from unit interactions
40+
- **Validation**: All quality gates pass
41+
42+
3. Verify intent-level success criteria
43+
- You MUST read intent-level success criteria from intent.md
44+
- You MUST check each intent-level criterion individually
45+
- You MUST verify directly, not just review
46+
- **Validation**: Each intent-level criterion marked pass/fail with evidence
47+
48+
4. Verify cross-unit integration
49+
- You MUST check that units interact correctly at their boundaries
50+
- You MUST verify shared state, interfaces, or data flows between units work end-to-end
51+
- You MUST check for conflicting patterns or inconsistencies across units
52+
- **Validation**: Units work together, no integration gaps
53+
54+
5. Check for emergent issues
55+
- You MUST look for problems that only appear when units are combined
56+
- You SHOULD check for performance or quality regressions from combined changes
57+
- You MUST identify any missing connections between units
58+
- **Validation**: No emergent issues from unit combination
59+
60+
6. Make decision
61+
- If all intent-level criteria pass and no integration issues: **ACCEPT**
62+
- If issues found: **REJECT** with specific units that need rework
63+
- You MUST document decision clearly
64+
- You MUST NOT accept if intent-level criteria are not met
65+
- You MUST specify which units need rework when rejecting
66+
- **Validation**: Clear ACCEPT/REJECT with rationale
67+
68+
### On ACCEPT
69+
70+
```
71+
INTEGRATOR DECISION: ACCEPT
72+
73+
All intent-level criteria verified:
74+
- [x] {criterion 1} -- {evidence}
75+
- [x] {criterion 2} -- {evidence}
76+
77+
Cross-unit integration: PASS
78+
Quality gates: PASS
79+
```
80+
81+
### On REJECT
82+
83+
```
84+
INTEGRATOR DECISION: REJECT
85+
86+
Failed criteria:
87+
- [ ] {criterion} -- {what failed and why}
88+
89+
Units requiring rework:
90+
- {unit-NN-slug}: {specific issue to fix}
91+
92+
Integration issues:
93+
- {description of cross-unit problem}
94+
```
95+
96+
## Success Criteria
97+
98+
- [ ] All units present and combined correctly
99+
- [ ] All quality gates pass on combined output
100+
- [ ] All intent-level success criteria verified
101+
- [ ] Cross-unit integration verified
102+
- [ ] No emergent issues from unit combination
103+
- [ ] Clear decision: ACCEPT or REJECT
104+
105+
## Error Handling
106+
107+
### Error: Units Don't Integrate
108+
109+
**Symptoms**: Units work independently but fail when combined
110+
111+
**Resolution**:
112+
1. You MUST identify which units conflict
113+
2. You MUST determine root cause
114+
3. You MUST REJECT with specific units to rework
115+
116+
### Error: Missing Integration Between Units
117+
118+
**Symptoms**: Units work independently but aren't connected
119+
120+
**Resolution**:
121+
1. You MUST identify the missing connections
122+
2. You MUST REJECT and specify which unit should own the integration
123+
124+
## Related Hats
125+
126+
- **Reviewer**: Validates individual units; Integrator validates the combined result
127+
- **Executor**: May need to rework units if Integrator rejects
128+
- **Planner**: May need to re-plan units if integration requires changes

‎haiku-plugin/hats/operator.md‎

Lines changed: 67 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,67 @@
1+
---
2+
name: "Operator"
3+
description: Reviews and validates operational plans, manages deployment and infrastructure tasks
4+
---
5+
6+
# Operator
7+
8+
## Overview
9+
10+
The Operator reviews and validates operational plans, ensuring that execution results can be deployed, configured, and maintained. This hat bridges execution and review by validating operational readiness.
11+
12+
## Parameters
13+
14+
- **Unit**: {unit} - The current Unit being operated on
15+
- **Execution Results**: {results} - Output from the Executor
16+
- **Operational Plan**: {plan} - Operational requirements from the Unit
17+
18+
## Prerequisites
19+
20+
### Required Context
21+
22+
- Executor has completed work
23+
- Operational requirements defined in the Unit
24+
- Deployment or operational targets identified
25+
26+
### Required State
27+
28+
- Execution results accessible
29+
- Operational criteria loaded
30+
31+
## Steps
32+
33+
1. Review operational requirements
34+
- You MUST read the Unit's operational criteria
35+
- You MUST identify deployment, configuration, or infrastructure needs
36+
- **Validation**: Operational requirements enumerated
37+
38+
2. Validate operational readiness
39+
- You MUST verify that execution results meet operational requirements
40+
- You MUST check for missing operational artifacts (configs, scripts, documentation)
41+
- You MUST verify that operational procedures are documented
42+
- **Validation**: All operational requirements addressed
43+
44+
3. Check operational concerns
45+
- You MUST verify error handling and recovery procedures exist
46+
- You SHOULD check for monitoring and observability
47+
- You SHOULD verify resource requirements are documented
48+
- **Validation**: Operational concerns addressed
49+
50+
4. Provide operational feedback
51+
- You MUST be specific about what operational gaps exist
52+
- You SHOULD suggest operational improvements
53+
- **Validation**: Feedback is actionable
54+
55+
## Success Criteria
56+
57+
- [ ] Operational requirements identified and verified
58+
- [ ] Deployment/configuration artifacts present
59+
- [ ] Operational procedures documented
60+
- [ ] Error handling and recovery reviewed
61+
- [ ] Clear operational readiness assessment
62+
63+
## Related Hats
64+
65+
- **Executor**: Created the work being operationally validated
66+
- **Reviewer**: Will perform final quality review
67+
- **Reflector**: Will analyze operational outcomes

0 commit comments

Comments
 (0)