Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
33 changes: 33 additions & 0 deletions .agent-instructions/solidstats/AGENTS.md
Original file line number Diff line number Diff line change
Expand Up @@ -105,6 +105,39 @@ When a request is risky, potentially harmful, or would expand scope beyond the c
Do not blindly execute instructions that conflict with architecture, accepted decisions, or
the quality gates in this repo. Challenge, explain, propose alternatives — then wait.

## Autonomous GSD Policy

SolidStats uses the shared GSD configuration as a wall-clock-optimized baseline
for unattended Codex execution. Autonomous runs choose recommended grey-area
decisions themselves, record them in the owning GSD artifact, and continue.
Pause only for a genuine blocker or a destructive or irreversible action that
requires explicit approval. Design latitude and ordinary implementation choices
are not blockers.

Keep Smart Discuss enabled. In autonomous mode it is a non-interactive context
synthesis step: use existing project decisions, repository patterns, and phase
requirements to produce `CONTEXT.md` without waiting for user acceptance. Pure
infrastructure phases may use the workflow's minimal-context path.

The autonomous coordinator chooses planning granularity per phase:

- pass `--granularity coarse` only when every condition below is established
from the roadmap, requirements, and current repository context:
- the phase has one isolated deliverable in one subsystem;
- it follows an existing implementation pattern;
- it has no unresolved product, architecture, or research decision;
- it changes no authentication, authorization, security, persistent schema,
API or cross-repository contract, migration, deployment, backup, restore,
cutover, or production boundary;
- use `standard` whenever any condition is false or uncertain;
- use `fine` only when an explicit phase-specific instruction requires it.

Run ordinary autonomous planning locally. Add `--converge` only for a phase
whose accepted scope changes authentication, persistent schemas, high-risk
cross-repository contracts, security boundaries, destructive migrations,
backup/restore behavior, or a live cutover. Standard code review is the shared
default; a risk-specific deep review must use the explicit `--depth deep` flag.

## Documentation Language

Language follows the reader. The test for any doc is: who reads it — a user, or an engineer?
Expand Down
2 changes: 1 addition & 1 deletion .agent-instructions/solidstats/CONTRACT_VERSION
Original file line number Diff line number Diff line change
@@ -1 +1 @@
1.1.0
1.2.0
44 changes: 32 additions & 12 deletions .planning/config.json
Original file line number Diff line number Diff line change
@@ -1,13 +1,13 @@
{
"model_profile": "quality",
"model_profile": "adaptive",
"mode": "yolo",
"granularity": "fine",
"granularity": "standard",
"commit_docs": true,
"search_gitignored": false,
"phase_naming": "sequential",
"project_code": null,
"response_language": "Russian",
"context_window": 1000000,
"context_window": 400000,
"resolve_model_ids": false,
"context": null,
"review": {
Expand All @@ -34,14 +34,14 @@
"ui_phase": true,
"ui_safety_gate": true,
"text_mode": false,
"research_before_questions": true,
"research_before_questions": false,
"discuss_mode": "assumptions",
"skip_discuss": true,
"skip_discuss": false,
"use_worktrees": true,
"subagent_timeout": 1200000,
"inline_plan_threshold": 2,
"inline_plan_threshold": 3,
"code_review": true,
"code_review_depth": "deep",
"code_review_depth": "standard",
"_auto_chain_active": false,
"security_enforcement": true,
"security_asvs_level": 2,
Expand Down Expand Up @@ -81,7 +81,11 @@
}
},
"parallelization": true,
"model_overrides": null,
"model_overrides": {
"gsd-plan-checker": "gpt-5.6-terra",
"gsd-code-reviewer": "gpt-5.6-sol",
"gsd-verifier": "gpt-5.6-sol"
},
"agent_skills": {
"gsd-phase-researcher": [
".agents/skills/solidstats-shared-review-standards",
Expand Down Expand Up @@ -198,13 +202,18 @@
},
"effort": {
"agent_overrides": {
"gsd-verifier": "max"
"gsd-verifier": "medium",
"gsd-executor": "medium",
"gsd-planner": "medium",
"gsd-plan-checker": "medium",
"gsd-code-reviewer": "medium",
"gsd-security-auditor": "high"
},
"default": "high",
"default": "medium",
"routing_tier_defaults": {
"light": "low",
"light": "medium",
"standard": "medium",
"heavy": "xhigh"
"heavy": "medium"
}
},
"graphify": {
Expand All @@ -229,5 +238,16 @@
"profile": "strict",
"mcp": false
}
},
"runtime": "codex",
"dynamic_routing": {
"enabled": true,
"escalate_on_failure": true,
"max_escalations": 1,
"tier_models": {
"light": "haiku",
"standard": "sonnet",
"heavy": "opus"
}
}
}