| name | sdlc-task |
| description | Run the full SDLC pipeline autonomously without human approval gates — all 7 roles execute, self-correct by looping back on discoveries, and produce a consolidated report. |
| argument-hint | [task description] |
SDLC Task Skill (Autonomous)
This skill runs the 7-gate SDLC pipeline without human approval gates. Read .agents/skills/sdlc-task/PIPELINE.md for the full pipeline process definition.
Arguments:
/sdlc-task <task description> — Run the full autonomous pipeline
SDLC Task Context
This skill produces the same artifacts as /sdlc — code, tests, inline documentation, and all gate reports — but without blocking on human approval at any gate.
- All 7 roles execute in sequence (Architect through Security Auditor)
- All gate artifacts are produced (ADR, SAR, Sprint Brief, Implementation Report, Code Review, Quality Report, Security Audit)
- Self-correction: discoveries that change architecture, scope, or security posture loop back to Gate 1 (max 2 cycles)
- Finding mitigation is driven by the invoker's directions — findings within scope are mitigated, findings outside scope are flagged
- Audit trail entries are marked
UNGATED
- Output: Consolidated Report summarizing all gates and re-evaluation cycles
Security of data and access controls:
- Security roles still execute their full analysis
- No interaction with any service or system may reduce the security posture of data passing through the process (REQ-5)
- External context files must pass a security review before loading (REQ-6)
