A free, open-source tool that scans your Git repositories for work that may qualify for Canadian SR&ED (Scientific Research and Experimental Development) tax credits.
Run it locally, in your own environment, on your own code. Nothing leaves your machine.
This skill analyses your Git commit history to identify patterns that suggest SR&ED-eligible R&D work:
- Iteration patterns — the same code being reworked repeatedly, suggesting systematic investigation
- Failed approaches — reverted commits and abandoned strategies, showing genuine uncertainty
- Strategy pivots — fundamental changes in technical approach, indicating the answer wasn't known in advance
- Experimental language — commit messages referencing hypotheses, experiments, benchmarks, and alternatives
It produces a markdown report classifying your work into "likely eligible," "possibly eligible," "support work," and "routine development" categories.
This is a preliminary scan, not a claim assessment. It uses the publicly available SR&ED eligibility criteria published by the CRA. It does not:
- Calculate qualified expenditures
- Draft T661 form narratives
- Assess how CRA reviewers would evaluate specific technical areas
- Provide the nuanced judgment that comes from decades of claim experience
For that, you need a specialist. Perly Consulting has over 25 years of SR&ED claim experience across hundreds of successful claims and CRA reviews.
This tool was built in a single Claude session, stripped down from the more advanced multi-agent toolset that Perly Consulting uses internally to prepare claims for clients.
Our production pipeline uses orchestrated Claude Code subagents running in parallel — preprocessing client documents, analysing multiple repositories simultaneously, cross-referencing task management systems, drafting T661 narratives to CRA word limits, and scoring eligibility against criteria refined over 25 years of actual CRA reviews. That pipeline incorporates proprietary knowledge about audit patterns, reviewer expectations, and claim-framing strategies that we've built up across hundreds of claims.
None of that proprietary IP is in this tool. What you're getting here is the public-domain skeleton: the CRA's published eligibility criteria applied to your Git history. It's genuinely useful as a first look — but it's a first look, not the full picture.
When we work with clients, our agents don't just scan repos. They integrate with your existing workflow:
- Task management — we connect to Linear, Jira, Asana, or whatever your team uses to pull issue histories, sprint records, and project timelines that corroborate the R&D narrative
- Communication archives — Slack threads, meeting notes, and design documents that capture the uncertainty and decision-making process as it happened
- Multi-repo orchestration — parallel analysis across your entire codebase, cross-referencing commits with task tickets and documentation
- T661 drafting pipeline — structured multi-stage writing, assessment, and compression to produce CRA-ready narratives within the strict word limits
- Expenditure mapping — tying eligible work to the people and costs that qualify for the tax credit
The result is a claim package built on evidence, not assertions — and framed by consultants who know what CRA reviewers look for because we've sat across the table from them for over two decades.
This skill uses the universal SKILL.md format and works across all major AI coding agents.
# Option 1: Copy into your project (team-shared)
mkdir -p .claude/skills/sred-eligibility-scan
cp SKILL.md .claude/skills/sred-eligibility-scan/
# Option 2: Copy to your user directory (personal, all projects)
mkdir -p ~/.claude/skills/sred-eligibility-scan
cp SKILL.md ~/.claude/skills/sred-eligibility-scan/Then invoke with:
/sred-eligibility-scan ./my-repo --period 2025-01-01:2025-12-31
# Option 1: Project-scoped
mkdir -p .cursor/skills/sred-eligibility-scan
cp SKILL.md .cursor/skills/sred-eligibility-scan/
# Option 2: User-scoped (all projects)
mkdir -p ~/.cursor/skills/sred-eligibility-scan
cp SKILL.md ~/.cursor/skills/sred-eligibility-scan/Then invoke via the / menu or by asking the agent to scan your repos for SR&ED eligibility.
The SKILL.md format is cross-compatible. Copy the skill folder into your agent's skill directory:
# Gemini CLI
mkdir -p .gemini/skills/sred-eligibility-scan
cp SKILL.md .gemini/skills/sred-eligibility-scan/
# Codex CLI
mkdir -p .codex/skills/sred-eligibility-scan
cp SKILL.md .codex/skills/sred-eligibility-scan/If your agent doesn't support the SKILL.md format natively, you can paste the contents of SKILL.md as a prompt and ask it to follow the instructions.
# Scan a single repo for the last 12 months
/sred-eligibility-scan ./my-project
# Scan multiple repos with a specific fiscal year
/sred-eligibility-scan ./api ./ml-pipeline ./data-service --period 2025-04-01:2026-03-31
# Scan the current directory
/sred-eligibility-scan .The tool will:
- Enumerate commits in the claim period
- Identify candidate R&D activity using commit message patterns and iteration analysis
- Deep-dive into the most promising areas by examining actual code changes
- Classify findings against the three SR&ED eligibility tests
- Generate a structured markdown report
The report categorises findings into four levels:
| Category | What It Means |
|---|---|
| Likely Eligible | All three SR&ED tests appear satisfied based on available evidence |
| Possibly Eligible | R&D signals are present but additional context is needed |
| Support Work | May qualify as SR&ED support if tied to an eligible area |
| Unlikely Eligible | Appears to be routine development or standard practice |
"Likely Eligible" does not mean "will be approved by CRA." It means the Git evidence shows patterns consistent with eligible work — but those same patterns also appear in work that is entirely ineligible. This tool cannot distinguish between the two. Whether a claim succeeds depends on narrative quality, evidence depth, expenditure calculations, and eligibility factors that this tool does not assess.
An automated scan can identify potential. Turning that potential into a defensible claim requires:
- CRA review experience — knowing how reviewers assess specific technical domains
- Narrative expertise — framing work in language that demonstrates eligibility without triggering audit flags
- Expenditure calculation — correctly identifying and documenting qualified salaries, materials, and contractor payments
- Audit preparedness — building a claim that stands up to detailed CRA technical review
Perly Consulting has been doing exactly this for over 25 years. We've prepared hundreds of claims and supported clients through CRA reviews. We know what works and what doesn't — not from theory, but from two decades of practice.
Get in touch for a complimentary initial consultation:
- Web: perlyconsulting.com
- Email: info@perlyconsulting.com
Issues and improvements welcome. This tool is intentionally limited to publicly available SR&ED criteria — it does not and will not incorporate proprietary claim-preparation methodologies.
MIT — use it however you like. If it helps you identify eligible work and you'd like help turning it into a claim, you know where to find us.