Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/saitarrun/devforge-ai/sdlc-implementgit clone --depth 1 https://github.com/saitarrun/Devforge-aiWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00029 | $0.01327 |
| Opus 5 | $0.00015 | $0.00664 |
| Sonnet 5 | $0.00006 | $0.00265 |
| Haiku 4.5 | $0.00003 | $0.00133 |
Grade A, and why
sdlc-implement scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 211 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/sdlc-implement — Standalone Issue Implementation
Implements a single Linear issue or free-form spec using fullstack-engineer + the ralph-loop skill for self-correcting retry logic. Works independently of the full /sdlc pipeline.
Usage
/sdlc-implement SAI-41 # Linear mode: reads issue from Linear
/sdlc-implement "add JWT refresh token logic" # Free-form mode: infers DoD from spec
STEP 1: Detect Input Mode
Inspect the argument:
- If it matches the pattern
[A-Z]+-[0-9]+(e.g.SAI-41,LIN-123) → Linear mode - Otherwise → Free-form mode
STEP 2: Read Spec + Extract DoD
Linear Mode
Use the Linear MCP to fetch the issue:
mcp__linear-server__get_issue({ id: "<ARGUMENT>" })
Extract from the response:
- Title — used as the commit subject
- Description — full context for the engineer
- Acceptance criteria — the checklist under "## Acceptance criteria" or "## DoD" becomes the Definition of Done
If no explicit AC section exists, infer DoD from the description: turn each "should", "must", or imperative statement into a checkbox.
Free-Form Mode
The argument string is the spec. Infer DoD by breaking the spec into verifiable outcomes:
- What file(s) should exist or change?
- What behaviour should be observable?
- What should typecheck + tests confirm?
Write the inferred DoD as a checklist (3–5 items minimum).
STEP 3: Load Ralph Loop Skill
Read skills/ralph-loop/SKILL.md to load the self-correcting loop methodology into context before invoking the engineer.
STEP 4: Invoke fullstack-engineer
Spawn the fullstack-engineer agent with the following context:
Spawn: Agent({
name: "fullstack-engineer",
description: "Implement: <title or spec summary>",
prompt: "
## Spec
<full issue description or free-form spec>
## Definition of Done
<DoD checklist extracted in Step 2>
## Instructions
Implement the spec above. Apply the ralph-loop skill:
### Inner Loop (apply per implementation attempt)
1. Run sentinel checks before verifying:
- Context drift: are you still addressing the full spec?
- Spec drift: are you writing code outside the declared scope?
2. Implement the changes.
3. Verification gateway:
- Run typecheck: `npm run typecheck` or `tsc --noEmit` (adapt to project)
- Run relevant tests: `npm test` or equivalent
4. If all checks pass → DoD satisfied, stop.
5. If checks fail → retry with fresh context:
- Re-read the spec + DoD
- Pin the failure output at the top of your reasoning
- Do NOT carry forward previous reasoning
- Max 2 retries. If still failing after 2 retries → stop and surface failure.
### Circuit Breaker
After 2 failed retries: report what failed, what you tried, and what the blocking error is.
Do not attempt further changes — surface to the user.
"
})
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 211 lines · 29 tokens per session scan A c3cb64c75b48
sdlc-implement is a command published in the GitHub repository saitarrun/Devforge-ai (5 stars, last pushed 19d ago), licensed Apache-2.0. It adds 29 tokens to every session and 1,327 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
team
The team command manages team members and messages for your Codev project. Team data is stored in codev/team/ and displayed in the Tower dashboard.
codev
The codev command manages project setup, maintenance, and framework-level operations.
tasks-to-ado
Convert existing tasks into actionable, dependency-ordered Azure DevOps work items for the feature based on available design artifacts.
sdlc-dev
Phase 3 — Development & Coding. Spawns 9 agents with 4 parallel groups: backend-engineer, frontend+mobile, database, fullstack-engineer.
sdlc-parallel
Run multiple agents in parallel across phases with shared context and collaboration.
sdlc-plan
Phase 1 planning — grill-me interview + 4 sequential agents (product-manager → business-analyst → software-architect → security-architect).