qraspi-research
Skill3 starsChanged 4 months agoMITView on GitHub
QRASPI Research phase -- map the solution LANDSCAPE for a new system, factual only, no recommendations. Use for "/qraspi-research <project>", "research the solution landscape for new X", "what libraries / prior art exist for X", "survey the options before architecting X". Mode switch: external-domain (no codebase) or inherited-repo. Do NOT use for QRSPI (an EXISTING codebase / adding a feature) -- that routes to qrspi-research. Do NOT use for the deprecated RPI workflow.
In this repository86 other files it gives its agents.
- AGENTS.mdAGENTS.md
- project-templates/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- project-templates/CLAUDE.mdCLAUDE.md
- alembic-migration-managerSkill
- architecture-journalSkill
- architecture-reviewSkill
- automated-code-reviewSkill
- axum-scaffolderSkill
- capture-consolidateSkill
- cargo-package-scaffoldSkill
- code-review-coachSkill
- confluence-guide-writerSkill
- dependency-mapperSkill
- doc-syncSkill
- dotnet-architecture-checklistSkill
- dotnet-controller-api-scaffolderSkill
- dotnet-security-reviewSkill
- dotnet-vertical-sliceSkill
- ef-migration-managerSkill
- email-captureSkill
- environment-healthSkill
- evaluate-testsSkill
- fastapi-scaffolderSkill
- fitness-functionsSkill
- jira-comment-writerSkill
- legacy-migration-analyzerSkill
- minimal-api-scaffolderSkill
- nuget-package-scaffoldSkill
- oss-vettingSkill
- pattern-tradeoff-analyzerSkill
- php-api-scaffolderSkill
- php-architecture-checklistSkill
- php-feature-sliceSkill
- php-migration-managerSkill
- php-package-scaffoldSkill
- php-security-reviewSkill
- pr-feedback-writerSkill
- pypi-package-scaffoldSkill
- python-architecture-checklistSkill
- python-feature-sliceSkill
- python-modernization-analyzerSkill
- python-security-reviewSkill
- qraspi-architectureSkill
- qraspi-graduateSkill
- qraspi-implementSkill
- qraspi-planSkill
- qraspi-questionsSkill
- qraspi-skeletonSkill
- qrspi-implementSkill
- qrspi-planSkill
- qrspi-questionsSkill
- qrspi-researchSkill
- qrspi-specSkill
- react-app-scaffolderSkill
- react-architecture-checklistSkill
- react-component-scaffolderSkill
- react-feature-sliceSkill
- react-modernization-analyzerSkill
- react-security-reviewSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 36
NO premature solution -- the state flag recommendations_made MUST stay false.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
QRASPI Research
"The map is not the territory." -- Alfred Korzybski
Core Philosophy
The Research phase produces a factual landscape map of the solution space -- never a recommendation. Greenfield has no ticket to hide and (in the default mode) no codebase to map; the failure mode shifts from QRSPI's "biased map" to premature solution -- the agent quietly picking FastAPI + Postgres before the Architecture phase has weighed a single trade-off. Recommendations are Architecture's job, gated behind ADRs. This phase catalogs what EXISTS in the solution space -- libraries, prior art, patterns, constraints -- and converts every comparative judgment into an open question for Architecture to decide.
The phase runs in one of two modes, detected at pre-flight (references/landscape-vs-codebase.md):
external-domain (default; pure greenfield) maps the problem domain and solution landscape via
research-synthesis and the web; inherited-repo (greenfield component inside an existing repo)
maps the host repo's conventions via the read-only research-* subagents.
Non-Negotiable Constraints:
- LANDSCAPE, not evaluation -- catalog what EXISTS in the solution space; convert every comparative judgment into an open question for Architecture. Recommendations are A's job.
- NO premature solution -- the state flag
recommendations_madeMUST stay false; do not pick a stack, framework, or library here - CITE every claim -- external-domain: source + credibility (via
research-synthesis); inherited-repo:file:line. An uncited claim is dropped. - MODE-AWARE firewall -- external-domain uses source-credibility discipline; inherited-repo passes ONLY a neutral topic string to the read-only subagents (never the project goal)
- CONTEXT BUDGET: keep utilization under 40%. At 60%, write
research.mdwith progress and tell the user to start a fresh session.
Workflow
PRE-FLIGHT
[ ] Locate the project folder thoughts/shared/qraspi/YYYY-MM-DD-{slug}/
[ ] Read the ANSWERED questions.md to scope the landscape (areas/constraints only)
-> If no questions.md exists, scope from the argument and note the gap
[ ] DETECT research_mode (see references/landscape-vs-codebase.md):
populated source tree at the target? no -> external-domain (default) yes -> inherited-repo
GATHER
external-domain: invoke research-synthesis; survey libraries, prior art, patterns, and
constraints via WebSearch/WebFetch; score source credibility; facts + citations
inherited-repo: derive a NEUTRAL topic string; spawn in PARALLEL via the Task tool, passing
ONLY that string:
@research-file-locator "Find all files related to: {neutral topic}"
@research-code-analyzer "Analyze the implementation of: {neutral topic}"
@research-pattern-finder "Find patterns and conventions related to: {neutral topic}"
wait for ALL THREE before synthesizing
NO subagent support (Pi): run the same three passes SEQUENTIALLY in this
session -- locate, then analyze, then patterns -- writing each pass to
research-{locate,analyze,pattern}.md before the next, and synthesizing from
those files. The ticket/goal must still never enter the session.
SYNTHESIZE
Organize into overview, landscape findings, options-on-the-table, constraints, open questions.
Convert EVERY comparative judgment ("X is faster than Y") into an open question for Architecture.
Compact to <= ~200 lines.
WRITE
thoughts/shared/qraspi/YYYY-MM-DD-{slug}/research.md (references/research-template.md)
Set status: complete
REPORT
Artifact path · 3-5 landscape facts · the options surfaced (NOT a pick) · open questions ·
"Review, then start a NEW session and run /qraspi-architecture"Exit criteria: research.md written as a factual landscape map; mode recorded; every claim
cited; every comparative judgment converted to an open question; recommendations_made: false;
user told to review before /qraspi-architecture.
State Block
<qraspi-research-state>
phase: PRE-FLIGHT | GATHER | SYNTHESIZE | WRITE | REPORT | COMPLETE
project_folder: thoughts/shared/qraspi/YYYY-MM-DD-{slug}/
research_mode: external-domain | inherited-repo
recommendations_made: false # MUST remain false -- the no-premature-solution firewall
neutral_topic: [ticket-free topic string | n/a for external-domain]
subagents_spawned: 0 | 1 | 2 | 3 # inherited-repo only
subagents_complete: 0 | 1 | 2 | 3 # inherited-repo only
open_questions: [count]
context_budget: under-40 | approaching-60 | checkpoint-now
status: in_progress | complete
</qraspi-research-state>Output Template
See references/research-template.md for the full research.md structure and frontmatter, and
references/landscape-vs-codebase.md for mode detection and the per-mode evidence rules.
Integration with Other Skills
| Skill | Relationship |
qraspi-questions | Prior phase. Its answered questions.md scopes the landscape. |
qraspi-architecture | Next phase. Consumes research.md; it -- not this phase -- makes the picks via ADRs. |
research-synthesis | The external-domain engine: source-credibility scoring and cross-referencing for the landscape map. |
qrspi-research | Brownfield sibling. Use it instead when the system EXISTS and you are mapping it to add a feature. |
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

