ariffazil/FORGE-context-compressor
skills/FORGE-context-compressor/SKILL.md
Compress oversized logs, transcripts, diffs, and telemetry before they exceed a host runtime context budget while preserving provenance and recovery pointers.
$ install --global
skillsauthnpx skillsauth add ariffazil/openclaw-workspace FORGE-context-compressorInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Jul 19, 2026, 3:45 AM162.7s1 file scanned
SKILL.md
- name:
- FORGE-context-compressor
- description:
- Compress oversized logs, transcripts, diffs, and telemetry before they exceed a host runtime context budget while preserving provenance and recovery pointers.
- forge_of:
- Kimi Code (FI-008) — EUREKA ZEN Phase 3 (gap fill · long-log compression for 256K-bound runtimes)
- forged:
- 2026-07-12T18:33:00Z
- rationale:
- Directive OPERATION EUREKA ZEN explicitly named "missing context-compression for long logs" as a Phase 3 forge candidate. Long-context runtimes (Grok 4.3 256K, Claude 1M variants) need a compression gate so they don't trip context-window boundaries mid-session. Phase 3 gap fill.
- binding:
- FORGE-* skills (cross-CLI), particularly for Grok (FI-010) and long-context adapters
- floor_scope:
- [F1, F2, F4, F8, F11, F13]
- tags:
- [forge, context-compression, long-log, 256k, rsi-breaker]
- status:
- NEW (Phase 3 gap fill)
FORGE · context-compressor
Cross-runtime compression gate for FORGE-* tools and AGI-* long-log producers. Used when raw input would exceed the host runtime's context budget.
When to invoke
- Raw log / transcript / diff > 80% of host runtime's context budget
- Multi-file audit where side-effects are dense
- Telemetry / observability feeds ingested in-session
- Pre-compaction: before
forge_execute_sealedif chain is dense
Compression contract
compress:
- strategy: rolling_window | semantic_chunks | bsdiff | paged_attention
chunk_size: tokens | lines | bytes
overlap_pct: number (0..50)
retention: verbatim | summary | pointer
provenance:
original_sha256: 12-char-prefix
compressor: <which>
captured_at: ISO-8601
Strategy selector
| Strategy | When |
|---|---|
| rolling_window | conversation stream — keep last N tokens verbatim, drop the rest to pointers |
| semantic_chunks | narrative + structured — chunk on discourse markers, embed each |
| bsdiff | code diffs — keep delta fields, drop noise (whitespace-only hunks) |
| paged_attention | mixed modal — paginate by scene/section, keep anchors |
RSI-breaker connection
If the compressor's output raises ΔS on re-decode:
- Drop to a more aggressive strategy (semantic_chunks → rolling_window)
- Emit
rsi_reason: "context_compression_too_aggressive"+ flag in next agent - Cap at 3 compressions per session — fail open to human after that
Grok 256K boundary
Grok Build 4.3 (FI-010) has 256K context. Pre-flight:
- Read raw ingest size via
forge_filesystem_stat. - If > 240K tokens (≈960KB text), enter compression gate.
- If > 256K tokens, emit
rsi_reason: "would_exceed_context"and HOLD.
Not instead of
Distinct from AGI-emd-encode (intake contract, witness readiness) and AGI-emd-metabolize (memory promotion). Compressor operates between observation and encode — it shrinks the substrate so encode can attach witnesses without exceeding context.
DITEMPA BUKAN DIBERI.
Related Skills
ariffazil/XAUUSD-trading-stack
development
VerifiedTrustedCommunity
Federation-wide gold (XAUUSD) trading capability. Python stack, OANDA broker, backtesting, macro signals, RSI strategy. Every organ has a role.
2SKILL.mdUpdated Jul 24, 2026
ariffazil/wealth-claim-state
development
VerifiedTrustedCommunity
Capital claim state management — tracks claim lifecycle across WEALTH organ.
2SKILL.mdUpdated Jul 24, 2026
ariffazil/warga-constitutional
development
VerifiedTrustedCommunity
Archived constitutional warga placeholder retained only for audit provenance. Do not use for active work; use the live arifOS governance and constitutional skills instead.
2SKILL.mdUpdated Jul 24, 2026
ariffazil/warga
testing
VerifiedTrustedCommunity
Warga (citizen) agent skills for AAA federation members. See subdirectories for specialized warga skills.
2SKILL.mdUpdated Jul 24, 2026