/** The local-report v4 dashboard shell (prototype), embedded verbatim. */ export declare const V4_TEMPLATE = "\n\n\n\n\n\naih report \u2014 harness health v0.5\n\n\n\n\n\n
\n
\n
\n \n
\n
\n

Enterprise AI Bootstrapping Harnessv0.5

\n
self-contained \u00B7 generated by aih report \u00B7 interpretive mode
\n
\n
\n
\n \n
\n UTC\u00B7\u00B7:\u00B7\u00B7:\u00B7\u00B7\n Local\u00B7\u00B7:\u00B7\u00B7:\u00B7\u00B7\n
\n \n \n
\n
\n\n\n \n \n \n \n \n \n \n\n\n\n\n\n
\n
\n \n DEMO DATA \u2014 illustrative figures for showcasing the report. Click the demo button to switch back.\n
\n\n
\n NEW\n v0.5 makes the report interpretive \u2014 anomaly engine flags what's unusual, forecasts predict budget burn, session replay traces agent behavior, and \u2318K jumps anywhere.\n \u00B7\n also new: token-flow Sankey, guardrail enforcement log\n
\n\n
\n\n
\n
\n \n
\n
\n
\n Harness health \u00B7 v0.5 snapshot \u00B7 interpretive\n

\n Platform is healthy,
\n three things want attention.\n

\n

\n MCP server up 99.2%, 1,847 calls today. The anomaly engine flagged 3 events: a cost spike on Tuesday, agent-stop hook degradation, and RULE_ROUTER.md drift. Forecast: context budget holds for 18 days at current burn.\n

\n
\n 93/100\n Mature\n
\n
\n
3anomalies2 warn \u00B7 1 bad
\n
18days to budgetat current burn
\n
1,847MCP callstoday
\n
$47.20cost / 7d\u221212% wow
\n
\n
\n
\n\n \n
\n
\n
\n
\n

Cost spike on Tuesday \u2014 opus-4 burn

\n high\n
\n

Tuesday's cost hit $12.40 \u2014 2.1\u00D7 the 30-day average. Traced to opus-4 usage on validate-guardrails (38 calls) after the guardrails-taxonomy edit. Cache miss rate climbed to 67%.

\n
\n
\n
$12.40Tue vs $5.80 avg
\n
\n
\n
\n
\n
\n

agent-stop hook degraded \u2014 2 failures

\n medium\n
\n

agent-stop hook success rate dropped to 91% this week (2 failures). Likely a stale path after the v0.4 upgrade. Other hooks firing at 100%. Usage-capture unaffected.

\n
\n
\n
91%23 fires \u00B7 2 fails
\n
\n
\n
\n
\n
\n

RULE_ROUTER.md drifted 2h ago

\n low\n
\n

Canon file ai-coding/RULE_ROUTER.md is +42 tokens out of sync with repo. Drift-check MCP tool detected it on the last post-commit. Other 4 tracked files still synced.

\n
\n
\n
+42 tokdrifted 2h ago
\n
\n
\n
\n\n
\n
Anomalies
3flagged
\n
MCP health
99.2%uptime
\n
Context budget
25.3%used
\n
Maturity
93/ 100
\n
Coherence
88%cross-CLI
\n
Days to budget
18forecast
\n
Cost / 7d
$47.20\u221212%
\n
\n
\n
\n
01flagship \u00B7 MCP
\n

Your harness is now an MCP server

\n

Twelve tools are exposed to four CLIs. 1,847 calls today, 99.2% uptime. The canon, code graph, and guardrails are now consumable by any MCP-aware agent \u2014 no copy-paste, no re-prompting.

\n
\n
\n
\n
\n
\n
\n
\n
\n
aih-mcptransport \u00B7 stdio + sse \u00B7 v0.5.1
\n live\n
\n
\n
1,847calls today+23% vs yesterday
\n
12tools exposedall canon-backed
\n
4consumersall connected
\n
\n
\n
Top tools \u00B7 last 24h12 registered
\n
search-canon428
\n
get-impact-radius312
\n
validate-guardrails287
\n
suggest-skill241
\n
next-task198
\n
drift-check156
\n
query-graph \u00B7 6 more225
\n
\n
\n
\n

Active consumers

\n
\n
claude743 calls
\n
codex512 calls
\n
cursor398 calls
\n
kiro194 calls
\n
\n
\n
Uptime \u00B7 24h99.2%
\n
\n
00:002 min downtime \u00B7 14:22now
\n
\n
\n
\n
\n
\n
\n
\n\n
\n
\n
02new in v0.5 \u00B7 signature
\n

Where the tokens actually go

\n

A token-flow Sankey: canon sources (left) \u2192 MCP tools that read them \u2192 CLIs that call those tools \u2192 models that consume the final prompts. Width is proportional to token volume. search-canon is the busiest artery.

\n
\n
\n
\n

Token flow \u00B7 today

10,122 tok routed
\n
\n
\n
\n canon sources\n MCP tools\n CLI consumers\n model endpoints\n 10,122 tokens \u00B7 1,847 tool calls \u00B7 4 CLIs \u00B7 5 models\n
\n
\n
\n
\n
\n\n
\n
\n
03new in v0.5
\n

Budget holds for 18 days \u2014 unless guardrails land

\n

Forecast models context-budget burn using a 14-day trend with 80% confidence band. At current burn (25.3% used), you hit 100% in 18 days. If the planned guardrails-taxonomy lands (+641 tok), that drops to 11 days.

\n
\n
\n
\n

Budget burn-down \u00B7 14d actual + 18d projected

80% confidence
\n
\n
\n 18 days\n to budget exhaustion\n at current burn \u00B7 +1.4% / day\n
\n
\n \n
\n
\n actual burn\n projected\n 80% confidence\n budget limit (40k tok)\n
\n
\n
18days to 100%at current burn
\n
11days if guardrails land+641 tok planned
\n
28days if canon pruned\u2212800 tok achievable
\n
\n
\n
\n
\n

What-if levers

\n
\n

Three levers change the forecast. Each is one command away.

\n
\n
\n
prune canon
\n
Trim 3 stale .aih.bak files
\n
+10 days \u00B7 aih prune --dry-run
\n
\n
\n
land guardrails
\n
Add guardrails-taxonomy.md
\n
\u22127 days \u00B7 aih guardrails --apply
\n
\n
\n
cache canon
\n
Enable MCP canon cache
\n
+5 days \u00B7 aih mcp --cache on
\n
\n
\n
\n
\n
\n
\n
\n
\n
0490 days
\n

Activity is accelerating \u2014 23-day streak

\n

Last 90 days of AI activity. 23 active days in the last 30, up from 14 in the prior 30. The MCP launch on day 67 is visible as a step-change in intensity.

\n
\n
\n
\n

Agent activity \u00B7 90 days

MCP launch \u00B7 day 67
\n
\n
\n
less
more
\n
\n
1,420total events
\n
63active days
\n
23current streak (days)
\n
longest streak \u00B7 31 days
\n
avg / active day \u00B7 22 events
\n
\n
\n
\n
\n
\n
\n\n
\n
\n
05cost economy
\n

$47.20 this week \u2014 down 12%

\n

Cost tracked per CLI via MCP tool-call attribution. Claude accounts for 68% ($32.10). Cache hit rate 41% saves ~$11/wk. Tuesday's anomaly (see above) accounts for 26% of the week's opus-4 spend.

\n
\n
\n
\n

Cost by CLI \u00B7 last 7 days

\u221212% wow
\n
\n
$47.20\u25BC $6.40 vs prior 7dSep 19 \u2014 Sep 25
\n
\n
share of spend$47.20 total
\n
\n
claude 68%
\n
codex 18%
\n
cursor 9%
\n
gem 3%
\n
kiro 2%
\n
\n
\n claude $32.10\n codex $8.50\n cursor $4.20\n gemini $1.40\n kiro $1.00\n
\n
\n
\n
daily burn \u00B7 30davg $6.74 / day
\n \n
\n
\n
\n
\n

Model routing

5 models
\n
\n

Haiku absorbs cheap repetitive calls; Opus is reserved for high-stakes reasoning. Tuesday's opus spike is visible.

\n
\n
share of token spend
\n
sonnet-4.552%
\n
haiku-4.531%
\n
opus-412%
\n
gpt-5-mini4%
\n
gemini-31%
\n
\n
\n \n Cache hit rate 41% \u2014 canon cached via MCP slashes repeat prompt cost by ~$11/wk.\n
\n
\n
\n
\n
\n\n
\n
\n
06skill ledger
\n

Six skills doing the heavy lifting

\n

Skill invocations logged via MCP. tdd-workflow leads at 142 calls, $18.40 saved via canon reuse. Three skills below ROI threshold \u2014 candidates for pruning.

\n
\n
\n
\n

Skill ledger \u00B7 last 7 days

11 skills \u00B7 444 invocations
\n
\n
TDtdd-workflowtesting
142$18.40
\n
CRcode-review-graphreview
89$11.20
\n
RPreact-patternsfrontend
67$7.80
\n
PTpython-testingtesting
45$5.40
\n
SRsecurity-reviewsecurity
38$6.10
\n
FPfastapi-patternsbackend
29$3.20
\n
CVcpp-coding-standardscpp
14$0.80
\n
LRlaravel-tddphp
11$0.60
\n
KTkotlin-ktorkotlin
9$0.40
\n
\n
\n
\n
\n\n
\n
\n
07coherence + drift
\n

CLIs agree on stack \u2014 one file drifted

\n

Cross-CLI coherence is 88%: 4 of 5 CLIs agree on stack, conventions, guardrails, adapters. Gemini diverges on test framework. RULE_ROUTER.md drifted 2h ago \u2014 see anomaly above.

\n
\n
\n
\n

Coherence matrix

88% agreement
\n
\n
\n
claude
codex
cursor
gemini
kiro
\n
Stack
\n
Conventions
\n
Guardrails
\n
Adapters
\n
Test fw
\n
\n
88%4 of 5 CLIs fully aligned with canon. Gemini diverges on guardrail strictness and test framework \u2014 review GEMINI.md adapter.
\n
\n
\n
\n

Drift detection

1 file drifted
\n
\n
\n
\n
Drift detectedRULE_ROUTER.md is out of sync with canon.
\n
2h ago
\n
\n
Tracked canon files
\n
\n
ai-coding/RULE_ROUTER.md+42 tokdrifted
\n
CLAUDE.md590 toksynced
\n
ai-coding/guardrails-taxonomy.md641 toksynced
\n
ai-coding/rules/agent-behavior-core.md565 toksynced
\n
ai-coding/command-policy.md1,322 toksynced
\n
\n
RULE_ROUTER.md drifted 2h ago. Run aih drift-check --apply to re-sync canon.
\n
\n
\n
\n
\n
\n
\n
08new in v0.5
\n

A session, replayed \u2014 claude \u00B7 09:18 \u2192 09:24

\n

One agent session traced end-to-end. Six minutes, nine MCP tool calls, two skill invocations, one commit. The filmstrip shows tool sequence and duration \u2014 search-canon opened the session, validate-guardrails closed it.

\n
\n
\n
\n

Session replay \u00B7 session-7c4a \u00B7 claude

09:18 \u2192 09:24 \u00B7 6m 12s
\n
\n
\n
\n
09:18:0009:20:0009:22:0009:24:12
\n
\n
\n search-canon 3\n get-impact-radius 2\n validate-guardrails 2\n suggest-skill 1\n drift-check 1\n thinking 2\n
\n
\n
9MCP tool calls
\n
2skill invocations
\n
1commit produced
\n
6m 12stotal duration
\n
\n
\n
\n
\n
\n\n
\n
\n
09new in v0.5
\n

Guardrails fired 47 times \u2014 4 blocked

\n

Guardrails aren't just a score anymore \u2014 every enforcement is logged. This week: 4 blocks (secrets, missing tests), 11 warnings, 3 bypasses (with reason), 29 allows. The 4 blocks caught 2 secret leaks before commit.

\n
\n
\n
\n

Guardrail enforcement log \u00B7 last 7 days

47 evaluations
\n
\n
\n
4blocked
\n
11warned
\n
3bypassed (with reason)
\n
29allowed
\n
\n
\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
TimeToolRuleActionReason
09:24 \u00B7 todayclaudesecrets-denyblockAPI key pattern detected in src/config.ts \u2014 commit halted, secret redacted
09:21 \u00B7 todayclaudevalidate-guardrailsallowAll 4 critical guardrails satisfied before commit
08:52 \u00B7 todayclaudetest-coveragewarnTest ratio 22.5% below 30% target \u2014 allowed with warning
17:44 \u00B7 yestcodexsecrets-denyblockAWS access key in .env.example \u2014 commit halted
17:30 \u00B7 yestcodexbranch-namingwarnBranch 'fix/telemetry' missing ticket ref \u2014 allowed
14:22 \u00B7 yestcursorimport-orderwarn3 imports out of canonical order \u2014 auto-fixed
11:08 \u00B7 yestclaudeno-console-logbypassDebug logging in dev-only module \u2014 explicit bypass with reason
09:15 \u00B7 yestkirotype-checkallowTypeScript strict mode passed, 0 errors
16:44 \u00B7 2dclaudesecrets-denyblockPrivate key block in test fixture \u2014 replaced with mock
10:30 \u00B7 2dcodextest-coveragewarnNew file src/utils.ts has no corresponding test \u2014 flagged
15:12 \u00B7 3dcursorbypass-requiredbypassForce-push to feature branch \u2014 admin bypass recorded
09:00 \u00B7 3dclaudesecrets-denyblockSlack webhook URL in deploy.sh \u2014 removed
\n
\n
\n
\n
\n
\n\n
\n
\n
10hook health
\n

Three of four hooks healthy

\n

pre-commit, post-commit, usage-capture firing at 100%. agent-stop at 91% \u2014 flagged as anomaly above. Likely a stale path after v0.4 upgrade.

\n
\n
\n
\n

Hook health \u00B7 last 7 days

3/4 healthy
\n
\n
pre-commit\u00B7 47 fires
47100%
\n
post-commit\u00B7 drift-check + usage-capture
47100%
\n
agent-stop\u00B7 23 fires \u00B7 degraded
2391%
\n
usage-capture\u00B7 universal post-commit
47100%
\n
agent-stop failed 2\u00D7 this week. Inspect .git/hooks/agent-stop \u2014 likely a stale path after the v0.4 upgrade.
\n
\n
\n
\n
\n\n
\n
\n
11adoption
\n

All wired except AGENTS.md

\n

8 of 9 adoption checks pass. Sole gap is AGENTS.md. 7 of 8 tools installed; 4 of 11 AI CLIs active via MCP.

\n
\n
\n

Tools installed

7/8
rgsgfdtreejqghcr-graphcomby
\n

Adoption checks

8/9
\u2713context-dir\u2713CLAUDE.md\u2713cursor-rules\u2713mcp\u2713gitleaks\u2713pre-commit\u2713claude-settings\u2713devcontainer\u2717AGENTS.md
\n

Repo status

feat/mcp-v05
  • mainmain2h ago
  • feat/mcp-v05+14\u221225m ago
  • fix/agent-stop+3\u221241d ago
uncommitted changes
\n

AI tooling \u00B7 MCP consumers

4/11 active \u00B7 MCP-enabled
claudecodexcursorkiroantigravitygeminicopilotwindsurfopencodezedkimi
\n
\n
\n\n
\n
\n
122,341 events
\n

Latest activity \u2014 MCP calls now dominant

\n

Since v0.4 launched, MCP calls are the leading event type, displacing raw commits. The replay above is one of these sessions.

\n
\n
\n

Event timeline

newest \u2192 oldest
session17:44codex
search-canon09:32mcp
skill09:18tdd (ecc)
impact-radius09:05mcp
commit08:52feat/mcp
validate-gr08:31mcp
+2,335 older events
\n
\n
\n\n
\n
\n
\n
\n \n
\n
\n
\n Harness health \u00B7 v0.5 snapshot \u00B7 interpretive\n

\n Regression caught
\n five anomalies, one root cause.\n

\n

\n Coherence dropped to 72%, drift detected, cost spiked +18%, two hooks degraded. The anomaly engine traced all five to one event: a guardrails-taxonomy edit on Tuesday that cascaded through the MCP tools.\n

\n
\n 79/100\n Emerging\n
\n
\n
5anomalies3 warn \u00B7 2 bad
\n
7days to budgetif uncorrected
\n
2,140MCP callstoday
\n
$84.30cost / 7d+18%
\n
\n
\n
\n\n
\n
\n

Cost spike \u2014 opus-4 on validate-guardrails

high
\n

Tuesday's cost hit $18.20 \u2014 2.7\u00D7 average. The guardrails-taxonomy edit triggered 67 validate-guardrails calls on opus-4, each re-reading the changed canon. Root cause of 3 other anomalies.

\n
$18.20Tue \u00B7 2.7\u00D7 avg
\n
\n
\n

Coherence dropped to 72%

high
\n

Two CLIs now diverge on guardrails and adapters. The taxonomy edit wasn't propagated to GEMINI.md and .cursor/rules. Root cause: drift-check ran but coherence-check didn't fire.

\n
72%was 88% \u00B7 \u221216pts
\n
\n
\n

Budget accelerated \u2014 7 days to exhaustion

medium
\n

Burn rate doubled since Tuesday. At current pace, budget exhausts in 7 days. The taxonomy edit added 1,840 tokens loaded per validate-guardrails call. Pruning canon buys back 10 days.

\n
7 dayswas 18 \u00B7 \u221211
\n
\n
\n\n
\n
Anomalies
5flagged
\n
MCP health
99.4%uptime
\n
Context budget
96%used
\n
Maturity
79/ 100
\n
Coherence
72%cross-CLI
\n
Days to budget
7forecast
\n
Cost / 7d
$84.30+18%
\n
\n\n
\n
\n
02new in v0.5 \u00B7 signature
\n

Token flow \u2014 the taxonomy edit is visible

\n

The Sankey shows the cascade: guardrails-taxonomy.md (left, thick) now feeds validate-guardrails disproportionately. opus-4 (right) is fatter than sonnet \u2014 the anomaly.

\n
\n

Token flow \u00B7 today

38,200 tok routed \u00B7 96% of budget
canon sourcesMCP toolsCLI consumersmodel endpoints
\n
\n\n
\n
\n
03forecast
\n

7 days to exhaustion \u2014 unless canon is pruned

\n

Burn rate doubled since Tuesday. Three levers: prune buys +10 days, cache buys +5, doing nothing loses the budget in 7.

\n
\n
\n

Budget burn-down \u00B7 14d actual + 7d projected

80% confidence
7 days\u25BC from 18 daysburn rate doubled \u00B7 +3.8% / day
actual burnprojected80% confidencebudget limit
7days to 100%at current burn
4if taxonomy stays+1,840 tok/call
17if canon prunedprune + cache
\n

Recommended action

One root cause, one fix: propagate the taxonomy edit to all adapters, then prune the stale .aih.bak files.
fix coherence
Propagate taxonomy to all 5 adapters
aih coherence --fix
prune canon
Remove 3 stale .aih.bak files
+10 days \u00B7 aih prune --apply
\n
\n
\n\n
\n
\n
08session replay
\n

The Tuesday session that started it

\n

Replay of the 14:22 session where the taxonomy edit cascaded. 11 validate-guardrails calls on opus-4 in one session \u2014 visible as the thick green band.

\n
\n

Session replay \u00B7 session-9f2e \u00B7 claude \u00B7 14:22

14:22 \u2192 14:31 \u00B7 9m 04s
14:22:0014:25:0014:28:0014:31:04
validate-guardrails 11search-canon 4impact-radius 2drift-check 3thinking 3
20MCP tool calls
11validate-guardrails
$12.40session cost
9m 04sduration
\n
\n\n
\n\n\n
\n
\n
\n \n \n esc\n
\n
\n
\u2191\u2193 navigate\u21B5 jump \u00B7 esc close
\n
\n
\n\n \n
\n\n\n\n";