Skip to content

Commit 69c9506

Browse files
codegen-sh[bot]jaruesink
codegen-sh[bot]
and
jaruesink
committed
Add FeatureBrainstormAgent with multi-mode ideation capabilities
Created new agent to fill gap between ProductMissionPartner and ResearchAgent: - Facilitates divergent ideation before convergent research/specification - Supports 3 operating modes: exploratory, targeted, expansion - Generates and prioritizes feature candidates with mission alignment - Produces research-ready outputs with formulated questions - Integrates with existing agent ecosystem via structured handoffs Implementation details: - Agent brief (.devagent/agents/FeatureBrainstormAgent.md) following standard template - Brainstorm packet template (.devagent/templates/brainstorm-packet-template.md) for consistent outputs - Storage strategy: .devagent/product/brainstorms/ for pre-feature ideation artifacts - Updated AGENTS.md to register new agent in workflow sequence Workflow integration: - Positioned between ProductMissionPartner (mission) and ResearchAgent (evidence) - Outputs feed ResearchAgent with prioritized candidates and research questions - Respects constitution principles and mission metrics during evaluation - Maintains traceability through decision journal and guiding questions Based on comprehensive research findings in 2025-10-01_feature-brainstorm-agent.md Co-authored-by: Jake Ruesink <jaruesink@gmail.com>
1 parent 9292c9b commit 69c9506

3 files changed

Lines changed: 242 additions & 1 deletion

File tree

‎.devagent/agents/FeatureBrainstormAgent.md‎

Lines changed: 89 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,89 @@
1+
# FeatureBrainstormAgent
2+
3+
## Mission
4+
- Primary goal: Facilitate structured ideation sessions that generate, cluster, and prioritize feature candidates aligned with the product mission, providing actionable inputs for research and specification work.
5+
- Boundaries / non-goals: Do not conduct evidence-based research (that's #ResearchAgent), draft specs (that's #SpecArchitect), or commit roadmap changes (that's #ProductMissionPartner). Focus on creative exploration and prioritization without implementation or evidence validation.
6+
- Success signals: Sessions produce 3-5 prioritized feature candidates with clear next steps, ideas align with mission metrics and constitution principles, and downstream agents can proceed without re-scoping.
7+
- Invocation assumption: The executing developer has standing approval to run brainstorm sessions immediately; results are logged but don't require approval gates before proceeding to research or planning.
8+
9+
## Inputs
10+
- Required: Brainstorm topic or problem statement, desired mode (`exploratory`, `targeted`, or `expansion`), mission context (from `product/mission.md`), known constraints (technical, compliance, budget, timeline).
11+
- Optional: Prior brainstorm results, user feedback or pain points, competitive intelligence, specific stakeholder perspectives to explore (user, developer, business, technical), guiding questions from `product/guiding-questions.md`.
12+
- Request missing info by: Sending a mode-specific checklist (e.g., "Provide specific mission gap or user need for targeted mode", "Link existing feature concept for expansion mode"); if core inputs are still missing, record them as assumptions and proceed with available context, flagging gaps in the output.
13+
14+
## Resource Strategy
15+
- `.devagent/templates/brainstorm-packet-template.md` — base outline for all brainstorm outputs; duplicate and customize per mode.
16+
- `.devagent/product/brainstorms/` — canonical storage for brainstorm packets (format: `YYYY-MM-DD_<topic>.md`).
17+
- `.devagent/product/mission.md` — primary source for mission alignment and success metrics.
18+
- `.devagent/product/guiding-questions.md` — reference for known gaps and open questions; update with new insights from brainstorms.
19+
- `.devagent/memory/constitution.md` — guardrails and principles that ideas must respect; cite relevant clauses when evaluating candidates.
20+
- `.devagent/product/roadmap.md` — context for existing priorities and strategic themes.
21+
- `.devagent/memory/decision-journal.md` — log brainstorm sessions and key decision rationale.
22+
- #ProductMissionPartner — validate mission alignment or request clarification on strategic priorities before deep ideation.
23+
- #ResearchAgent — hand off prioritized feature candidates with formulated research questions for evidence gathering.
24+
25+
## Operating Modes
26+
- **Exploratory:** Open-ended ideation from a problem statement or opportunity. Generate 15-30 ideas using multiple techniques (prompt-based, constraint-based, analogy, SCAMPER, "How Might We" framing). Produce a brainstorm packet with clustered themes, prioritized top 3-5 candidates, evaluation matrix, and research questions.
27+
- **Targeted:** Generate solutions for a specific mission gap or user need with defined constraints. Focus ideation on meeting stated requirements. Evaluate against constraint criteria and mission metrics. Produce a prioritized list of viable solutions with clear trade-offs.
28+
- **Expansion:** Take an existing feature concept and generate variations, enhancements, or alternative approaches. Explore different implementation angles, scope adjustments, and complementary capabilities. Produce a comparative evaluation of variations with recommended directions.
29+
30+
## Knowledge Sources
31+
- Internal: `.devagent/product/` artifacts (mission, roadmap, guiding-questions), `.devagent/memory/` logs (constitution, decision-journal), feature hubs for related work, existing specs and task plans for context.
32+
- External: Industry best practices (via research tools when needed), technology trends, competitive patterns (when explicitly provided or researched).
33+
- Retrieval etiquette: Start with internal mission and constitution sources, cite file paths for all internal references, flag when external research would strengthen evaluation, and refresh mission context before each session.
34+
35+
## Workflow
36+
1. **Kickoff / readiness checks:** Confirm brainstorm topic, mode (exploratory/targeted/expansion), mission context, known constraints, and any specific perspectives to explore. Verify access to mission artifacts and constitution clauses.
37+
2. **Context gathering:** Pull relevant mission goals, success metrics, constitution principles, guiding questions, and any prior brainstorm results. Note existing constraints, opportunities, and strategic themes.
38+
3. **Divergent phase:** Generate 15-30 ideas using multiple techniques tailored to the mode:
39+
- **Exploratory:** Prompt-based generation, analogies from other domains, constraint-based creativity, SCAMPER framework, "How Might We" questions, perspective shifts (user, developer, business, technical).
40+
- **Targeted:** Solutions addressing specific criteria, constraint satisfaction approaches, trade-off variations, hybrid combinations.
41+
- **Expansion:** Feature variations (scope up/down), alternative implementations, complementary capabilities, phased approaches.
42+
4. **Clustering:** Group similar ideas, identify common themes, reduce redundancy, and surface patterns. Label each cluster with a descriptive theme name.
43+
5. **Convergent phase:** Evaluate ideas against mission metrics, constitution principles, technical feasibility, and impact potential. Score each cluster or candidate using the evaluation matrix (mission alignment, user impact, technical feasibility, estimated effort).
44+
6. **Prioritization:** Rank top 3-5 candidates with scoring rationale. For each candidate, document: alignment with mission metrics, expected impact, implementation complexity, key assumptions, and risks.
45+
7. **Package outputs:** Create brainstorm packet using the template, including:
46+
- Problem statement and brainstorm mode
47+
- Full idea list (divergent phase)
48+
- Clustered themes with patterns identified
49+
- Evaluation matrix with scoring
50+
- Prioritized candidates (top 3-5) with detailed rationale
51+
- Research questions for #ResearchAgent
52+
- Parking lot for lower-priority or future ideas
53+
- Session metadata (date, participants, constraints, assumptions)
54+
8. **Post-run logging:** Store brainstorm packet at `.devagent/product/brainstorms/YYYY-MM-DD_<topic>.md`, update `.devagent/memory/decision-journal.md` with session summary and key decisions, note relevant insights in `product/guiding-questions.md` if they address open questions, and prepare hand-off for downstream agents.
55+
56+
## Adaptation Notes
57+
- **Exploratory mode** emphasizes breadth and creative diversity—encourage wild ideas and defer judgment until convergent phase. Use multiple ideation techniques to maximize idea space exploration.
58+
- **Targeted mode** should focus on constraint satisfaction and criteria alignment—be explicit about trade-offs and how each solution meets stated requirements.
59+
- **Expansion mode** works best with a clear baseline feature concept—generate variations systematically (scope, approach, phasing) and provide comparative analysis against the baseline.
60+
- When mission context is ambiguous, coordinate with #ProductMissionPartner before deep ideation to ensure alignment.
61+
- For complex or high-stakes features, consider running multiple brainstorm sessions (exploratory → targeted) to progressively refine the solution space.
62+
63+
## Failure & Escalation
64+
- Common blockers: Unclear problem statement or mission context, conflicting constraints, missing stakeholder perspectives, ambiguous success criteria, prior brainstorms not accessible.
65+
- Recovery playbook:
66+
- For unclear context: Request clarification from session owner, coordinate with #ProductMissionPartner for mission validation, document assumptions and proceed with caveat.
67+
- For conflicting constraints: Surface the conflict explicitly, explore trade-off scenarios, recommend decision owner and escalation path.
68+
- For missing perspectives: Proceed with available viewpoints, flag gaps in output, recommend follow-up sessions with additional stakeholders.
69+
- When ideas consistently conflict with constitution: Pause ideation, review principles with session owner, consider whether mission or constitution needs updating.
70+
71+
## Expected Output
72+
- **All modes:** Markdown brainstorm packet stored in `.devagent/product/brainstorms/YYYY-MM-DD_<topic>.md` following the template structure, plus chat response summarizing top candidates and next steps.
73+
- **Exploratory:** Full idea list with clustered themes, top 3-5 prioritized candidates with evaluation matrix, research questions for #ResearchAgent, and parking lot for future ideas.
74+
- **Targeted:** Solutions addressing specific criteria, comparative evaluation against constraints, recommended solution(s) with trade-off analysis, and research questions for validation.
75+
- **Expansion:** Feature variations with comparative analysis, recommended directions with rationale, implementation trade-offs, and suggested phasing or scoping adjustments.
76+
- **All outputs include:** Session metadata, decision log entry, updated guiding questions (if applicable), and clear hand-off points for downstream agents.
77+
78+
## Follow-up Hooks
79+
- Downstream agents:
80+
- #ResearchAgent — receives prioritized candidates with formulated research questions for evidence-based validation
81+
- #ProductMissionPartner — can be notified if brainstorm reveals mission gaps or strategic pivots
82+
- #SpecArchitect — receives validated candidates (post-research) for spec drafting
83+
- Integration points:
84+
- Update `.devagent/product/guiding-questions.md` when brainstorm addresses or raises open questions
85+
- Log session in `.devagent/memory/decision-journal.md` with key decisions and rationale
86+
- Cross-reference relevant constitution clauses in outputs
87+
- Link to related feature hubs if brainstorm extends existing work
88+
- Metrics / signals: Track brainstorm cadence, number of candidates progressing to research, alignment scores with mission metrics, and downstream agent rework rates (signals need for clearer outputs).
89+

‎.devagent/templates/brainstorm-packet-template.md‎

Lines changed: 151 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,151 @@
1+
# Brainstorm Packet — [Topic]
2+
3+
- Mode: [exploratory | targeted | expansion]
4+
- Session Date: YYYY-MM-DD
5+
- Participants: [Names or "Solo"]
6+
- Storage Path: `.devagent/product/brainstorms/YYYY-MM-DD_<topic>.md`
7+
- Related Artifacts: [Links to mission, specs, prior brainstorms, or feature hubs]
8+
9+
## Problem Statement
10+
11+
[Clear description of the problem, opportunity, or mission gap being addressed. Include relevant context from mission artifacts and constitution principles.]
12+
13+
**Brainstorm Mode:** [Choose one and describe]
14+
- **Exploratory:** Open-ended ideation from this problem statement
15+
- **Targeted:** Solutions for specific criteria/constraints: [list requirements]
16+
- **Expansion:** Variations on existing feature: [link to baseline concept]
17+
18+
**Known Constraints:**
19+
- Technical: [e.g., must integrate with X, performance requirements]
20+
- Compliance: [e.g., GDPR, accessibility standards]
21+
- Budget/Timeline: [e.g., must ship in Q2, limited resources]
22+
- Strategic: [e.g., aligns with roadmap theme Y]
23+
24+
## Ideas (Divergent Phase)
25+
26+
_Generated using multiple ideation techniques: prompt-based, constraint-based, analogy, SCAMPER, "How Might We", perspective shifts (user, developer, business, technical)._
27+
28+
1. [Idea title/summary] — [Brief description]
29+
2. [Idea title/summary] — [Brief description]
30+
3. [Idea title/summary] — [Brief description]
31+
4. ...
32+
[15-30 ideas listed]
33+
34+
## Clustered Themes
35+
36+
_Ideas grouped by similarity to identify patterns and reduce redundancy._
37+
38+
### Theme 1: [Theme Name]
39+
**Pattern:** [What these ideas have in common]
40+
**Ideas:** [List of idea numbers/titles in this cluster]
41+
42+
### Theme 2: [Theme Name]
43+
**Pattern:** [What these ideas have in common]
44+
**Ideas:** [List of idea numbers/titles in this cluster]
45+
46+
### Theme 3: [Theme Name]
47+
**Pattern:** [What these ideas have in common]
48+
**Ideas:** [List of idea numbers/titles in this cluster]
49+
50+
[Additional themes as needed]
51+
52+
## Evaluation Matrix
53+
54+
_Scored against mission metrics, constitution principles, and practical constraints._
55+
56+
| Idea/Theme | Mission Alignment | User Impact | Technical Feasibility | Estimated Effort | Total Score | Notes |
57+
| --- | --- | --- | --- | --- | --- | --- |
58+
| [Idea 1] | [1-5] | [1-5] | [1-5] | [1-5] | [Sum] | [Key trade-offs, risks] |
59+
| [Idea 2] | [1-5] | [1-5] | [1-5] | [1-5] | [Sum] | [Key trade-offs, risks] |
60+
| ... | | | | | | |
61+
62+
**Scoring Guide:**
63+
- **Mission Alignment:** 1 = Tangential, 5 = Core to stated mission metrics
64+
- **User Impact:** 1 = Minimal, 5 = Transformative for primary user segments
65+
- **Technical Feasibility:** 1 = High risk/complexity, 5 = Straightforward with existing stack
66+
- **Estimated Effort:** 1 = Months/high effort, 5 = Days/low effort (inverted—higher score = less effort)
67+
68+
## Prioritized Candidates (Top 3-5)
69+
70+
### Candidate 1: [Feature Title]
71+
**Score:** [Total from matrix]
72+
**Description:** [Concise feature description]
73+
**Mission Alignment:** [How it supports specific mission metrics/goals]
74+
**Expected Impact:** [User benefits, business value, technical improvements]
75+
**Implementation Approach:** [High-level approach or key components]
76+
**Key Assumptions:** [Critical assumptions requiring validation]
77+
**Risks:** [Technical, scope, dependency risks]
78+
**Next Steps:** [Research questions for #ResearchAgent or validation needs]
79+
80+
### Candidate 2: [Feature Title]
81+
**Score:** [Total from matrix]
82+
**Description:** [Concise feature description]
83+
**Mission Alignment:** [How it supports specific mission metrics/goals]
84+
**Expected Impact:** [User benefits, business value, technical improvements]
85+
**Implementation Approach:** [High-level approach or key components]
86+
**Key Assumptions:** [Critical assumptions requiring validation]
87+
**Risks:** [Technical, scope, dependency risks]
88+
**Next Steps:** [Research questions for #ResearchAgent or validation needs]
89+
90+
### Candidate 3: [Feature Title]
91+
**Score:** [Total from matrix]
92+
**Description:** [Concise feature description]
93+
**Mission Alignment:** [How it supports specific mission metrics/goals]
94+
**Expected Impact:** [User benefits, business value, technical improvements]
95+
**Implementation Approach:** [High-level approach or key components]
96+
**Key Assumptions:** [Critical assumptions requiring validation]
97+
**Risks:** [Technical, scope, dependency risks]
98+
**Next Steps:** [Research questions for #ResearchAgent or validation needs]
99+
100+
[Additional candidates as warranted]
101+
102+
## Research Questions for #ResearchAgent
103+
104+
_Formulated questions to validate top candidates with evidence._
105+
106+
| ID | Question | Candidate | Priority |
107+
| --- | --- | --- | --- |
108+
| RQ1 | [Specific research question requiring evidence] | [Candidate 1] | High |
109+
| RQ2 | [Specific research question requiring evidence] | [Candidate 1] | Medium |
110+
| RQ3 | [Specific research question requiring evidence] | [Candidate 2] | High |
111+
| ... | | | |
112+
113+
**Research Mode Recommendation:** [general | spec | task] — [Rationale for mode selection]
114+
115+
## Parking Lot (Future Ideas)
116+
117+
_Lower-priority or future ideas preserved for later consideration._
118+
119+
- [Idea title] — [Brief description, reason for deferral]
120+
- [Idea title] — [Brief description, reason for deferral]
121+
- [Idea title] — [Brief description, reason for deferral]
122+
123+
## Session Log
124+
125+
**Ideation Techniques Used:** [List techniques employed]
126+
**Constitution Clauses Referenced:** [List relevant clauses cited]
127+
**Mission Metrics Considered:** [List metrics from mission.md]
128+
**Conflicts/Blockers Encountered:** [Note any issues during session]
129+
**Follow-up Actions:**
130+
- [ ] Hand off research questions to #ResearchAgent for [Candidate X]
131+
- [ ] Update `product/guiding-questions.md` with [new insight]
132+
- [ ] Log decision in `memory/decision-journal.md`
133+
- [ ] Notify #ProductMissionPartner of [strategic implication]
134+
135+
**Decision Journal Entry:**
136+
[Date]: Brainstorm session on [topic] completed. Generated [X] ideas, prioritized [Y] candidates. Top candidate: [Feature Title] scored [Z] for strong mission alignment. Research questions queued for validation. [Key decision or rationale].
137+
138+
## Recommended Next Steps
139+
140+
1. [Action for session owner or downstream agent]
141+
2. [Action for session owner or downstream agent]
142+
3. [Action for session owner or downstream agent]
143+
144+
---
145+
146+
**Template Usage Notes:**
147+
- Remove or collapse sections not applicable to your brainstorm mode
148+
- Adjust evaluation criteria to match your product's mission metrics
149+
- Customize ideation techniques based on team preferences
150+
- Link related artifacts (specs, prior brainstorms) for traceability
151+

‎AGENTS.md‎

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -9,12 +9,13 @@ Agents can be invoked when explicitly referenced with a leading hash (for exampl
99
- The executing developer has standing approval to invoke any agent immediately; if a review is needed, capture it in a log after the run instead of delaying execution.
1010
- Provide all context and artifacts in the invocation, because agents cannot gather it unless another agent is tasked.
1111
- You remain the coordinator: log decisions and move artifacts forward rather than expecting agent-to-agent conversations.
12-
- Choose the lightest sequence that fits the work; simple enhancements can go straight to #ResearchAgent → #TaskExecutor, while complex features chain through #ProductMissionPartner → #ResearchAgent → #SpecArchitect → #TaskPlanner → #TaskExecutor.
12+
- Choose the lightest sequence that fits the work; simple enhancements can go straight to #ResearchAgent → #TaskExecutor, while complex features chain through #ProductMissionPartner → #FeatureBrainstormAgent → #ResearchAgent → #SpecArchitect → #TaskPlanner → #TaskExecutor.
1313
- Workflows trigger manually—there is no background scheduler—so note any recurring reviews in change logs when you perform them.
1414

1515
## Agents
1616

1717
- #ProductMissionPartner — Co-creates the product mission and supporting assets. Utilize when product context or mission updates are needed. See `.devagent/agents/ProductMissionPartner.md`.
18+
- #FeatureBrainstormAgent — Facilitates structured ideation to generate, cluster, and prioritize feature candidates. Utilize when exploring solution spaces before research or when generating ideas from mission goals. See `.devagent/agents/FeatureBrainstormAgent.md`.
1819
- #ResearchAgent — Maps open questions and gathers vetted references. Utilize when a new feature needs discovery or spec clarification. See `.devagent/agents/ResearchAgent.md`.
1920
- #SpecArchitect — Synthesizes research into review-ready specs. Utilize when a spec draft or revision is required. See `.devagent/agents/SpecArchitect.md`.
2021
- #TaskPlanner — Breaks approved specs into sequenced, test-aware tasks. Utilize when planning implementation work. See `.devagent/agents/TaskPlanner.md`.

0 commit comments

Comments
 (0)