-
Notifications
You must be signed in to change notification settings - Fork 1
Adding a Vendor
Process for adding a new agent CLI to the SpecRoute supported matrix. Adding a vendor is a new column, not a fork. The framework's content (PRDs, specs, prompts, rules) stays unchanged; the runtime shell adapts.
The add-vendor skill walks through this interactively — see .claude/skills/add-vendor/SKILL.md.
runtimes/.<vendor>/
├── README.md per-runtime setup steps (model on existing READMEs)
├── settings.template.json or .toml — whatever the vendor consumes
├── (agents/) only if the vendor has an agent primitive
├── (skills/) only if the vendor has a skill primitive
├── (commands/ or commands.json) only if the vendor has commands
└── (hooks/) if the vendor has hooks
Pattern: only ship the directories the vendor actually consumes. Don't add empty agents/ if the vendor has no agent concept.
The matrix appears in three files. Update all three in lock-step:
-
README.md— supported vendor matrix section. -
AGENTS.md— supported vendor matrix section. -
agentic-docs/agent-cli-integrations.md— supported vendor matrix section.
Each row: Vendor | Runtime dir | Root context file | Skills | Agents | Commands | Hooks | MCP config.
Use – for what the vendor doesn't support. If hooks are supported, also update the per-vendor hook depth table in agentic-docs/agent-cli-integrations.md and the event matrix in hooks/README.md.
rules/<vendor>-rules.md
Document:
- How the vendor surfaces rules (always-on file, glob-matched, etc.).
- How the shared rules (
engineering,code-review,security,documentation) map onto the vendor's format. - Vendor-specific frontmatter conventions.
- Known limitations vs Claude / Codex.
If the vendor consumes MCP servers:
- Document the vendor's MCP config shape and path.
- Add
runtimes/mcp/render/render_<vendor>.py. - Update
runtimes/mcp/servers.yamlper-serverenabled_forlists to include the new vendor.
See MCP Integration.
If the vendor shares artifact shapes with Claude or Codex:
-
For agents — extend
tools/sync-skills.pyto include the new vendor's agents directory. - For skills — same.
If the shapes are genuinely different (Kiro's .kiro.hook vs Claude's hooks.json), no sync target; document the asymmetry.
For each vendor that supports hooks:
hooks/<vendor>/
├── hooks.template.json (or .toml or per-hook files)
└── scripts/ optional supporting shell scripts
Use the existing per-vendor templates under hooks/ as the model. Document the event list and blocking semantics in hooks/README.md.
If the vendor reads a specific filename at session start (like Claude's CLAUDE.md), add a short delegation shim at the repo root:
# <VENDOR>.md
<Vendor> shim for this repository. Read `AGENTS.md` first; it is the canonical
vendor-neutral source of truth.
## <Vendor>-Specific Context
- (Vendor-specific runtime location, frontmatter quirks, hook config path…)
## Before Publishing
(Vendor-specific sanitization or pre-commit notes.)Length: 30–50 lines. See Multi-Vendor Context Files.
After the changes, run /audit — it checks vendor-matrix consistency across the three files and flags broken cross-references.
-
runtimes/.<vendor>/exists with a README and the appropriate template files. - Matrix row is identical across
README.md,AGENTS.md,agentic-docs/agent-cli-integrations.md. -
rules/<vendor>-rules.mddocuments how rules surface. - If the vendor consumes MCP: renderer exists,
servers.yamlper-serverenabled_forupdated. - If the vendor has hooks:
hooks/<vendor>/templates exist, event list documented inhooks/README.md. - If the vendor auto-loads a root file: delegation shim exists at repo root (30–50 lines).
- Per-runtime README walks through setup (rename templates, .gitignore additions, etc.).
-
/auditpasses.
Vendor onboarding is owned by the runtime-architect agent.
- Vendor Matrix — the contract
- Agent CLI Integrations — how each existing vendor is wired
- Cross-Vendor Sync — keeping things aligned
- Multi-Vendor Context Files — the delegation-shim pattern
Repository · Issues · Roadmap · Changelog · Security · License (Apache 2.0) — © Enovatr Labs
- Philosophy
- Spec-Driven Development
- Agentic Coding Model
- Automation Decision Framework
- Multi-Agent Orchestration
- Two-Tier Docs Pattern
- Multi-Vendor Context Files
- Documentation Structure
- Agent Memory
- Artifact Taxonomy
- PRDs
- Specs
- Agents
- Skills
- Commands
- Hooks
- Prompts
- Rules
- Frontmatter Contracts
- Sanitization