Agent Configuration & Fitness
Agent Context & Working Memory
Two-tier architecture: assembled messages inside the model active context window vs. reachable resources on disk. Source in UTC, displayed in MSK.
CAPACITY UNKNOWN
Assembly fit: unavailable
Source instructions: unavailable
History: unavailable
latest
Capped Prompt Load
0 capped chars
~0 est. tokens under builder cap
Actual System Message: unavailable
Context Budget Cap
unavailable
recorded cap unavailable
Remaining Headroom
n/a
margin before overflow trim
Tier 2 Reachable Knowledge
~1246.3 KB
242 files reachable via read_file
Prompt Budget & Fit
Prompt Budget Utilization: unavailable (recorded cap unavailable)active cycle
TRUNCATION/DROP STREAK: NO DATA ledger unreadable or no
phase: system_prompt rows recorded — not the same as a healthy 0, healthy: 0 consecutive cycles.WINDOW PRESSURE (24H): WITHIN BUDGET (p99 79.8%) healthy: ≤ 80% • p99 over 628 row(s) with a known context_window • window unknown: 177 rows (excluded, not counted as 0%) of 805 total in the last 24h.
Prompt Fit Event Telemetry
Published runtime summary · latest prompt row: 2026-09-30T17:08:43.188946Z · source: published
Rungfull
Dropped0 sections / 0 chars
Trimmed0 sections / 0 chars
Recent rows with drops0 / 25
Recent rows with trims0 / 25
Event window25/25 rows · newest_system_prompt_rows · 90d · 2026-09-28T06:13:40.792174Z → 2026-09-30T17:08:43.188946Z
Dropped section namesnone
Trimmed section namesnone
Dialogue Window: model window 98,304 tokens (executor llm_calls.context_window) · compaction reserve unknown (not published by runtime) · usable budget unknown · occupancy: 31,525 prompt tokens (executor) · present
Prompt Assembly & Context Architecture
TIER 1
In Active Context (Attention Window)
Assembled into system and user messages.
… 0 × "
---
" Separators0c
msghistory + tool resultsmodel window 98,304 tokens (executor llm_calls.context_window) · compaction reserve unknown (not published by runtime) · usable budget unknown · occupancy 31,525 prompt tokens (executor)
userruntime_context + task0c
➔
On-Demand Access
read_file · search_memory · exec
Tier 1 index costs: catalogue 0c · memory 0c · exec path unavailableTIER 2
Reachable on Disk (Zero Base Context)
Full instructions, lessons, and memory records.
Tier 1: Assembled Context Blocks (Recorded Order) format: unavailable
Blocks follow the ledger row's own recorded `sections` order for this cycle -- never a hard-coded assembly order. Click to view exact text.
Arithmetic Character Reconciliation
sections: unavailablesections breakdown: unavailable — This cycle row was recorded prior to structured section logging (#1379). Recorded total chars: 0. Per honesty rules, section sizes are not reconstructed.
Rule Owners static (until harness publishes rule_owners)
| Rule (fingerprint test) | Owning block | Owner class |
|---|---|---|
skip | agents | instance |
surface | agents | instance |
branch | agents | instance |
runner | agents | instance |
identity | identity | release |
budget | operating | release |
final_json | operating | release |
Reachable Knowledge & Access Paths
Tier 2: Reachable On-Demand Knowledge Base
Assets residing on disk, accessible by tool calls during cycle loop. Total: ~1246.3 KB across 242 files.
Skills catalogue: budget unavailablec · retained unavailable/unavailable · unavailable · omitted: none
Memory index: status unavailable · resident matched unavailable · resident missing: unavailable
Skill source join: catalogue · Tier 2 disk · skill fitness reads
| Skill | In catalogue | On disk | Reads | Confirmed | State |
|---|---|---|---|---|---|
| batch-grep | unavailable | yes | 7 | 5 | working |
| bounded-reading-search-prevent-context-bloat | unavailable | yes | 4 | 4 | working |
| chained-python-execution | unavailable | yes | 4 | 3 | working |
| check-target-completion | unavailable | yes | 5 | 4 | working |
| classify-python-module | unavailable | yes | 1 | 0 | working |
| command-output-bounding-prevent-token-exhaustion | unavailable | yes | 2 | 2 | working |
| composite-context-inspection | unavailable | yes | 1 | 0 | working |
| demand-and-reflection-lookup-paths | unavailable | yes | 4 | 4 | working |
| early-implementation-transition | unavailable | yes | 4 | 3 | working |
| early-skip-evaluation | unavailable | yes | 4 | 4 | working |
| early-turn-direct-file-inspection-and-editing-turns-1-4 | unavailable | yes | 2 | 2 | working |
| early-turn-staging | unavailable | yes | 5 | 4 | working |
| early-validation-and-commit-budgeting | unavailable | yes | 3 | 1 | working |
| eeebot-agent-work-review | unavailable | yes | 5 | 3 | working |
| fast-path-redundancy-skip | unavailable | yes | 14 | 12 | working |
| handle-loop-breaker | unavailable | yes | 5 | 3 | working |
| immediate-skip-protocol | unavailable | no | 2 | 0 | not on disk |
| incremental-script-refactoring-under-step-limits | unavailable | yes | 2 | 0 | working |
| inspect-verify-and-proof | unavailable | yes | 6 | 1 | working |
| jsonl-stream-filter | unavailable | yes | 3 | 3 | working |
| locate-failure-logs | unavailable | yes | 4 | 1 | working |
| memory-lookup | unavailable | yes | 5 | 4 | working |
| multi-script-refactor | unavailable | yes | 4 | 4 | working |
| pair-agents-instructions-with-tests | unavailable | yes | 8 | 8 | working |
| posix-background-execution-in-sh | unavailable | yes | 2 | 2 | working |
| pure-documentation-changes-test-bypass-workflow | unavailable | yes | 3 | 3 | working |
| raw-json-final-turn | unavailable | yes | 2 | 1 | working |
| recursive-unittest-testsuite-traversal | unavailable | yes | 2 | 1 | working |
| run-tests | unavailable | yes | 59 | 52 | working |
| search-subagent-archive | unavailable | yes | 1 | 1 | working |
| secret-verification-scoping | unavailable | yes | 3 | 1 | working |
| standard-test-runner-unittest-over-pytest | unavailable | no | 4 | 4 | not on disk |
| sync-agents-sections-in-structural-tests | unavailable | yes | 2 | 1 | working |
| target-git-log-history-inspection-single-file-fixes | unavailable | yes | 5 | 4 | working |
| targeted-file-slice | unavailable | yes | 4 | 1 | working |
| targeted-test-discovery | unavailable | no | 2 | 2 | not on disk |
| targeted-test-validation-legacy-glob-suites | unavailable | yes | 2 | 2 | working |
| turn-budget-checkpoints-80-iteration-cycles | unavailable | yes | 2 | 2 | working |
| verify-lessons-integrity | unavailable | yes | 9 | 6 | working |
Skills Store (36 skills • ~227.5 KB)
Corpus: present; indexed in Tier 1 viaskills_catalogue (0 chars)batch-grep10,087 B
View SKILL.md (10,087 bytes)
text on the LAN site only (0 chars)
bounded-reading-search-prevent-context-bloat1,106 B
View SKILL.md (1,106 bytes)
text on the LAN site only (0 chars)
chained-python-execution7,635 B
View SKILL.md (7,635 bytes)
text on the LAN site only (0 chars)
check-target-completion5,633 B
View SKILL.md (5,633 bytes)
text on the LAN site only (0 chars)
classify-python-module9,222 B
View SKILL.md (9,222 bytes)
text on the LAN site only (0 chars)
command-output-bounding-prevent-token-exhaustion1,803 B
View SKILL.md (1,803 bytes)
text on the LAN site only (0 chars)
composite-context-inspection4,326 B
View SKILL.md (4,326 bytes)
text on the LAN site only (0 chars)
demand-and-reflection-lookup-paths1,372 B
View SKILL.md (1,372 bytes)
text on the LAN site only (0 chars)
early-implementation-transition4,345 B
View SKILL.md (4,345 bytes)
text on the LAN site only (0 chars)
early-skip-evaluation758 B
View SKILL.md (758 bytes)
text on the LAN site only (0 chars)
early-turn-direct-file-inspection-and-editing-turns-1-42,148 B
View SKILL.md (2,148 bytes)
text on the LAN site only (0 chars)
early-turn-staging8,723 B
View SKILL.md (8,723 bytes)
text on the LAN site only (0 chars)
early-validation-and-commit-budgeting14,062 B
View SKILL.md (14,062 bytes)
text on the LAN site only (0 chars)
eeebot-agent-work-review11,270 B
View SKILL.md (11,270 bytes)
text on the LAN site only (0 chars)
fast-path-redundancy-skip6,440 B
View SKILL.md (6,440 bytes)
text on the LAN site only (0 chars)
handle-loop-breaker10,603 B
View SKILL.md (10,603 bytes)
text on the LAN site only (0 chars)
incremental-script-refactoring-under-step-limits14,216 B
View SKILL.md (14,216 bytes)
text on the LAN site only (0 chars)
inspect-verify-and-proof17,168 B
View SKILL.md (17,168 bytes)
text on the LAN site only (0 chars)
jsonl-stream-filter1,544 B
View SKILL.md (1,544 bytes)
text on the LAN site only (0 chars)
locate-failure-logs5,622 B
View SKILL.md (5,622 bytes)
text on the LAN site only (0 chars)
memory-lookup4,747 B
View SKILL.md (4,747 bytes)
text on the LAN site only (0 chars)
multi-script-refactor3,308 B
View SKILL.md (3,308 bytes)
text on the LAN site only (0 chars)
pair-agents-instructions-with-tests4,628 B
View SKILL.md (4,628 bytes)
text on the LAN site only (0 chars)
posix-background-execution-in-sh4,593 B
View SKILL.md (4,593 bytes)
text on the LAN site only (0 chars)
pure-documentation-changes-test-bypass-workflow1,338 B
View SKILL.md (1,338 bytes)
text on the LAN site only (0 chars)
raw-json-final-turn10,342 B
View SKILL.md (10,342 bytes)
text on the LAN site only (0 chars)
recursive-unittest-testsuite-traversal9,273 B
View SKILL.md (9,273 bytes)
text on the LAN site only (0 chars)
run-tests8,463 B
View SKILL.md (8,463 bytes)
text on the LAN site only (0 chars)
search-subagent-archive5,992 B
View SKILL.md (5,992 bytes)
text on the LAN site only (0 chars)
secret-verification-scoping1,745 B
View SKILL.md (1,745 bytes)
text on the LAN site only (0 chars)
sync-agents-sections-in-structural-tests13,443 B
View SKILL.md (13,443 bytes)
text on the LAN site only (0 chars)
target-git-log-history-inspection-single-file-fixes5,122 B
View SKILL.md (5,122 bytes)
text on the LAN site only (0 chars)
targeted-file-slice11,988 B
View SKILL.md (11,988 bytes)
text on the LAN site only (0 chars)
targeted-test-validation-legacy-glob-suites1,309 B
View SKILL.md (1,309 bytes)
text on the LAN site only (0 chars)
turn-budget-checkpoints-80-iteration-cycles947 B
View SKILL.md (947 bytes)
text on the LAN site only (0 chars)
verify-lessons-integrity7,622 B
View SKILL.md (7,622 bytes)
text on the LAN site only (0 chars)
Lessons Corpus
corpus: PRESENT lessons/index.md: PRESENTCorpus Files: 144 lesson records on disk (~579.2 KB total):
lessons/README.md (1,798 B)
lessons/adjacent_test_suite_verification.md (5,559 B)
lessons/already_implemented_skip_emission.md (3,766 B)
lessons/assigned_demand_dedup_protocol.md (3,248 B)
lessons/avoid_bare_time_in_exec.md (5,164 B)
lessons/avoid_bundled_test_executions.md (3,072 B)
lessons/avoid_full_program_cprofile.md (4,474 B)
lessons/avoid_full_repo_pytest_collection.md (4,496 B)
lessons/avoid_module_constants_in_default_args.md (3,708 B)
lessons/avoid_state_report_inspection.md (5,331 B)
lessons/avoiding_repeat_failures.md (6,767 B)
lessons/avoiding_repeated_failures.md (2,093 B)
lessons/baseline_validator_checks.md (4,573 B)
lessons/batch-consecutive-file-edits.md (4,463 B)
lessons/batch_companion_file_edits.md (4,961 B)
lessons/batched_test_append_redirection.md (4,359 B)
lessons/blocked_file_remediation.md (3,511 B)
lessons/bounded_initial_reconnaissance.md (2,979 B)
lessons/bounded_inspection_turns.md (4,142 B)
lessons/bounded_script_reads.md (4,341 B)
lessons/canonical_cli_assertions.md (4,818 B)
lessons/chained_git_commit_workflow.md (2,572 B)
lessons/check_existing_tests_before_proposals.md (3,060 B)
lessons/chunked_file_reading.md (3,965 B)
lessons/cli_deprecation_json_stdout.md (4,902 B)
lessons/cli_subprocess_exit_codes.md (2,803 B)
lessons/cli_verification_state_isolation.md (5,564 B)
lessons/cohesive_import_edits.md (4,443 B)
lessons/compact_script_enhancement.md (4,666 B)
lessons/companion_cycle_test_verification.md (3,726 B)
+114 more lessons...
Working Memory Store
corpus: PRESENT memory/index.md: PRESENTIndexed via Tier 1 memory block. 62 files on disk (~439.6 KB total):
memory/HISTORY.md (53 B)
memory/MEMORY.md (196 B)
memory/MEMORY_ARCHIVE.md (5,150 B)
memory/archive/MEMORY-legacy.md (94,613 B)
memory/archive/done-stubs.md (28,337 B)
memory/archive/legacy-complete.md (95,660 B)
memory/auxiliary_tool_call_discipline.md (2,388 B)
memory/blocked_task_early_exit.md (5,865 B)
memory/compound_task_scoping.md (5,066 B)
memory/confirmation_status.json (7,099 B)
memory/direct_edit_discipline.md (4,842 B)
memory/facts/agents-instruction-assertions.md (321 B)
memory/facts/base-test-failure-isolation.md (429 B)
memory/facts/bounded-text-inspection.md (427 B)
memory/facts/cli-positional-optional-for-standalone-flags.md (359 B)
memory/facts/cli-stderr-json-separation.md (304 B)
memory/facts/cli-test-suite-timeout.md (423 B)
memory/facts/daemon-runtime-paths.md (433 B)
memory/facts/decay_archival_policy.md (2,094 B)
memory/facts/do-not-touch.md (1,987 B)
memory/facts/doc-code-block-placeholder-normalization.md (374 B)
memory/facts/doc-snippet-syntax-line-offset.md (333 B)
memory/facts/error-card-gitignore-handling.md (417 B)
memory/facts/feed-cache-preflight-gate.md (421 B)
memory/facts/git-database-permissions.md (89 B)
memory/facts/host-metrics-heartbeat-timestamp.md (313 B)
memory/facts/identity.md (363 B)
memory/facts/key-paths.md (475 B)
memory/facts/markdown-catalog-link-integrity.md (450 B)
memory/facts/markdown-structural-testing.md (411 B)
+32 more files...
AGENTS.md Charter
AGENTS.md charter (149 lines)
text on the LAN site only (7,644 chars)
Goals Charter
goals charter absent
Skill Fitness (Reads & Usage)
| Skill | Reads (skill_fitness/reads.json) | Confirmed Usage (skill_fitness/reads.json; skill_fitness/evals.jsonl when present) |
|---|---|---|
| run-tests | 59 | 52 confirmed |
| fast-path-redundancy-skip | 14 | 12 confirmed |
| verify-lessons-integrity | 9 | 6 confirmed |
| pair-agents-instructions-with-tests | 8 | 8 confirmed |
| batch-grep | 7 | 5 confirmed eval delta: +3 |
| inspect-verify-and-proof | 6 | 1 confirmed |
| eeebot-agent-work-review | 5 | 3 confirmed |
| memory-lookup | 5 | 4 confirmed eval delta: +5 |
| early-turn-staging | 5 | 4 confirmed |
| handle-loop-breaker | 5 | 3 confirmed |
| target-git-log-history-inspection-single-file-fixes | 5 | 4 confirmed |
| check-target-completion | 5 | 4 confirmed |
| multi-script-refactor | 4 | 4 confirmed |
| chained-python-execution | 4 | 3 confirmed |
| early-skip-evaluation | 4 | 4 confirmed |
| standard-test-runner-unittest-over-pytest | 4 | 4 confirmed |
| early-implementation-transition | 4 | 3 confirmed |
| bounded-reading-search-prevent-context-bloat | 4 | 4 confirmed |
| demand-and-reflection-lookup-paths | 4 | 4 confirmed |
| targeted-file-slice | 4 | 1 confirmed |
| locate-failure-logs | 4 | 1 confirmed |
| pure-documentation-changes-test-bypass-workflow | 3 | 3 confirmed |
| early-validation-and-commit-budgeting | 3 | 1 confirmed |
| secret-verification-scoping | 3 | 1 confirmed |
| jsonl-stream-filter | 3 | 3 confirmed |
| targeted-test-discovery | 2 | 2 confirmed |
| incremental-script-refactoring-under-step-limits | 2 | 0 confirmed |
| early-turn-direct-file-inspection-and-editing-turns-1-4 | 2 | 2 confirmed |
| turn-budget-checkpoints-80-iteration-cycles | 2 | 2 confirmed |
| command-output-bounding-prevent-token-exhaustion | 2 | 2 confirmed |
| posix-background-execution-in-sh | 2 | 2 confirmed |
| recursive-unittest-testsuite-traversal | 2 | 1 confirmed |
| targeted-test-validation-legacy-glob-suites | 2 | 2 confirmed |
| sync-agents-sections-in-structural-tests | 2 | 1 confirmed |
| immediate-skip-protocol | 2 | 0 confirmed |
| raw-json-final-turn | 2 | 1 confirmed |
| composite-context-inspection | 1 | 0 confirmed |
| classify-python-module | 1 | 0 confirmed |
| search-subagent-archive | 1 | 1 confirmed |
Proposer
last proposal: Implement should_backoff_demand in scripts/demand_backoff.py
model: an/gemini-3.8-flash-high · 259 calls in the last 7 days
| Day | Calls | Tokens | Mean latency |
|---|---|---|---|
| 2026-09-30 | 61 | 589.7K | 18s |
| 2026-09-28 | 42 | 361.1K | 15s |
| 2026-09-27 | 43 | 398.6K | 16s |
| 2026-09-26 | 42 | 393.4K | 16s |
| 2026-09-25 | 27 | 232.3K | 16s |
| 2026-09-24 | 44 | 470.7K | 23s |