# eeebot / tech-tree

integrations96
confirmed integration ratio52.9%
repeat failure rate · scorecard13.6%
old repeat failure rate · keep 7d13.6%
execution failures35 events · 27 tasks · 10.1%
model unavailable (known)0 events · 0 tasks · 0.0%
model call incompleteno data
failure cause unknownno data
self_dedup rejections7 events · 1 tasks · 2.0%
repeat failure rate · new21.7% · новая формула без self_dedup (#1765); не улучшение работы
tokens / integration2.1M
held-out4/4
supplier-paused0 cycle(s) / 0s
Window: 7d (complete)
data: 30s old · generated 22:56 UTC

Now / Active Focus

Health: DEGRADED last integrated cycle is 2.2d old
Direction: automated-structural-assertion-for-doc-o active lever: loop.confirmed_integration_ratio · aim: higher
Latest Cycle: cycle-584bcce25fb7 (cycle-584bcce25fb7 [6420733c])
Next Up: 5 ranked items, published 2026-09-30T22:39:37.628050Z

Head: self-derived · vector V1 — provenance(operator<self-derived), then vector(V1<V2), as demand._priority_items

  1. priority self-derived Check runner compatibility vector V1
  2. priority self-derived Backoff chronic duplicate demand IDs vector V1
  3. priority self-derived Filter linter diagnostics by diff lines vector V1
  4. priority self-derived Rewrite bare time exec command vector V1
  5. priority self-derived Detect pipeline exit code check vector V1

Ranked by provenance (operator < self-derived), then vector. Selection (eeebot#1708/#1711) takes the eligible operator item first, else LRU rotation (#902) -- the presented order above may differ from what is picked next.

Charter:

Operator Charter

source: release_goals_md

> Immutable. Ships in the release tree; the gate rejects any cycle that touches it. ## Goal statement eeebot is a resource-aware, self-evolving autonomous agent on an old, slow eeepc host. Its purpose, set by the operator, is ordered: Vector 1 is the primary goal; Vector 2 is secondary; and every day owes something to the screen (Vector 3). ## Vector 1 (PRIMARY) — Self-Improvement of the Agent's Own Work Make the instance's own tooling, knowledge and workflows better at running improvement cycles. Its object is what the loop can itself commit: `scripts/`, `skills/`, `tests/`, `lessons/`, `memory/`, `docs/` and `surfaces/` inside `eeebot-self-evolving/`. That means sharpening the tools the cycle actually uses; working more precisely with them; mining the ledger, results and lessons for what worked and what failed and turning that into applied changes; retiring what no longer earns its keep; and optimizing what it owns for this hardware, from better algorithms to caching. Every optimization claim must come with a before/after measurement. ### Operator-executed, and not counted as loop progress The harness in `ozand/eeebot` — prompt assembly, compaction, the gate, demand collection, the proposer — is outside the loop's commit surface, and so are `systemd/`, `ops/` and `state/`. Work there is real, and it is the operator's to implement. The loop's route is a proposal under `docs/` carrying the before/after measurement that justifies it; the numbers are what make it count. Modules in Rust, C++ or C are that kind of proposal, never a commit. ## Vector 2 (SECONDARY) — Operator Interface and Process Transparency Give the operator transparent insight into what the bot is doing, and interfaces to track and steer work. On this host terminal rendering is usually the most efficient medium (pixel-art output such as images/eeebot.png); a speed-optimized local page is also valid. An interface counts only if the operator can use it and that use can be observed; abandoned one

Self-Derived Priorities

  • 62 vector V1 ·
  • 63 vector V1 ·
  • 64 vector V1 ·
  • 65 vector V1 ·
  • 66 vector V1 ·
  • 67 vector V1 ·
  • 68 vector V1 ·
  • 69 vector V1 ·
  • 70 vector V1 ·
  • 71 vector V1 ·
Bridge Exit Streak: 0 failures (crash_record) does not see SIGTERM kills (eeebot#1683)
Feed Freshness: heldout: 1.3h/12.0h host_metrics: 1.1h/24.0h llm_calls: 5m/24.0h usage: 1.4h/12.0h validator_harness_parent: 2.5h/12.0h
Doc Budget Guard: unavailable
Strategist: ran 2026-09-22T00:00:10 — inputs 5/5, 1 hypotheses, 2 advisories
GitHub CI: read 2026-09-30T22:56:20.017395Z
  • ozand/eeebot: enabled=unanswerable, runs_old, success, last 2026-09-29 20:27:26 MSK
  • ozand/eeebot-self-evolving: enabled=unanswerable, recent, failure, last 2026-10-01 01:35:51 MSK
  • ozand/eeebot-ops-dashboard: enabled=unanswerable, recent, success, last 2026-10-01 01:50:36 MSK
Systemd drift: ✓ present 12 defect(s), scanned 2026-09-30 02:00:01 MSK
  • installed, not in release: (owner: unknown)
  • installed, not in release: (owner: unknown)
  • installed, not in release: (owner: unknown)
  • installed, not in release: (owner: unknown)
  • installed, not in release: (owner: unknown)
  • installed, not in release: (owner: dashboard)
  • installed, not in release: (owner: preset)
Local CI: ✓ present pass: passed
Executor Model: ✓ present openai/un/qwen3.8-27b-gguf (self_hosted), 543 recent calls checked
Artifact Graph: 248 artifacts 57 ready 191 leaves
oldest leaves: scripts/analyze_cycle_duration.py, scripts/analyze_pass_streak.py, scripts/backlog_health.py, scripts/benchmark_api_latency.py, scripts/benchmark_cpu.py
Recent Failed Bridge Exits (192)
  • 2026-09-29T17:58:13.176038Z [status TERM]:
  • 2026-09-28T19:13:08.035979Z [status TERM]:
  • 2026-09-27T03:36:07.449464Z [status TERM]:
  • 2026-09-26T23:46:42.693760Z [status TERM]:
  • 2026-09-26T01:15:38.564656Z [status TERM]:
  • 2026-09-25T20:21:33.346570Z [status TERM]:
  • 2026-09-25T17:46:26.784884Z [status TERM]:
  • 2026-09-25T16:48:32.054022Z [status TERM]:
  • 2026-09-25T15:50:38.057494Z [status TERM]:
  • 2026-09-25T13:05:15.387276Z [status TERM]:
Served / completed history
Served: goal-gap ×2 defect ×10 priority ×5 other ×11
Completed: none

Explore

Cycle Feed (Recent 3)