substitute-eraser
npx skills add https://github.com/plurigrid/asi --skill substitute-eraser
Agent 安装分布
Skill 文档
Substitute Eraser
Scan existing codebases for placeholder tokens. Generate remediation plan.
Purpose
Audit existing files for substitution tokens (TODO, FIXME, placeholder, mock, etc.) and produce actionable remediation plan. Distinct from accept-no-substitutes which validates agent output.
Scope: Existing Files
This skill scans what already exists:
- Source code files
- Configuration files
- Documentation with stale placeholders
- Test fixtures that leaked into production
Complements accept-no-substitutes (output validation).
Trit Assignment
- Trit: -1 (MINUS/VALIDATOR)
- Hue: 270° (violet – deep scan)
- Role: Codebase auditor, technical debt detector
Scan Workflow
1. Discovery
# Scan current directory
just substitute-scan .
# Scan specific path
just substitute-scan src/
2. Classification
| Severity | Tokens | Action |
|---|---|---|
| CRITICAL | TODO, FIXME, placeholder, xxx | Must fix before merge |
| WARNING | mock-, fake-, stub-* (outside tests) | Review context |
| INFO | example_, demo_ | Document or remove |
3. Remediation Report
Output format:
SUBSTITUTE ERASER REPORT
========================
Scanned: 142 files
Found: 23 substitutions
CRITICAL (7):
src/auth.py:42 TODO: implement token refresh
src/api.py:118 placeholder value
src/db.py:55 FIXME: race condition
...
WARNING (12):
src/service.py:30 mock_client (not in test file)
...
INFO (4):
README.md:15 example_config
...
REMEDIATION PLAN:
1. [CRITICAL] src/auth.py:42 - Implement token refresh logic
2. [CRITICAL] src/api.py:118 - Replace placeholder with actual value
...
Context-Aware Exceptions
Acceptable Locations
| Pattern | Acceptable In |
|---|---|
mock-*, fake-*, stub-* |
*_test.py, test_*.py, /tests/ |
example_* |
README.md, /docs/, /examples/ |
demo_* |
/demo/, documentation |
TODO |
Issue tracker references with ID |
Exception Syntax
Mark intentional placeholders:
# SUBSTITUTE-OK: mock used for test isolation
mock_client = MockHTTPClient()
Commands
# Full scan with report
just substitute-scan <path>
# Critical only (CI mode)
just substitute-critical <path>
# Generate remediation tasks
just substitute-tasks <path> --output=github # GitHub issues
just substitute-tasks <path> --output=linear # Linear tickets
just substitute-tasks <path> --output=todo # TODO file
# Interactive fix mode
just substitute-fix <path>
Integration with GF(3)
Operates as MINUS (-1) in audit triads:
substitute-eraser(-1) + code-generator(+1) + review-coordinator(0) = 0
Emits rejection signal when scan finds violations above threshold.
Additional Resources
Reference Files
references/patterns.md– Detection regex patterns (shared with accept-no-substitutes)references/remediation.md– Fix strategies per token type
Scripts
scripts/scan.py– Main scanning scriptscripts/report.py– Report generation
SDF Interleaving
This skill connects to Software Design for Flexibility (Hanson & Sussman, 2021):
Primary Chapter: 10. Adventure Game Example
Concepts: autonomous agent, game, synthesis
GF(3) Balanced Triad
substitute-eraser (+) + SDF.Ch10 (+) + [balancer] (+) = 0
Skill Trit: 1 (PLUS – generation)
Secondary Chapters
- Ch6: Layering
- Ch4: Pattern Matching
- Ch2: Domain-Specific Languages
Connection Pattern
Adventure games synthesize techniques. This skill integrates multiple patterns.