docs: [#849] address review feedback on validator docs

- validators.md: drop "session rollup" claim from telemetry section
  (only per-validator events are emitted today); make `checked` /
  `concurrency_limit` optional in the result-shape schema to match
  what the validators actually return on the no-models path
- dbt-tools.md: mention both opt-in flags (ENABLED + SHADOW) and the
  zero-overhead default
- skills.md: correct the auto-load placement docs (prepended BEFORE
  the available-skills listing, not appended after — placement was
  deliberate)
- benchmark/ade-bench/README.md: add 'text' language to the directory
  tree code fence for markdownlint MD040

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: anandgupta42

benchmark/ade-bench/README.md

@@ -6,7 +6,7 @@ It deliberately does **not** ship the trace files, the per-trial result JSONs, t
 
 ## What's in this folder
 
-```
+```text
 benchmark/ade-bench/
 ├── README.md                              ← you are here
 ├── altimate_code_agent/                   ← drop-in agent module for ade-bench

docs/docs/configure/skills.md

@@ -62,10 +62,12 @@ description: ...
 ---
 ```
 
-At session start, after the standard `<available_skills>` listing, every matched
-skill body is appended to the system prompt under:
+At session start, every matched skill body is prepended to the system prompt
+(BEFORE the standard `<available_skills>` listing — placement matters: putting
+the auto-loaded block first frames the bodies as binding "rules of the road"
+rather than background reference) under:
 
-```
+```xml
 <auto_loaded_skill name="<skill-name>">
 ... full skill body ...
 </auto_loaded_skill>

docs/docs/data-engineering/tools/dbt-tools.md

@@ -333,6 +333,11 @@ validators** that fire automatically after the agent declares done. They run
 `altimate-dbt test` and `altimate-dbt schema-verify` against every model
 modified during the session and block "done" if anything failed.
 
-This is **opt-in** today (`ALTIMATE_VALIDATORS_ENABLED=1`) — see the
-[Validators page](../validators.md) for the full reference, env var
-catalogue, performance characteristics, and the phased rollout plan.
+This is **opt-in** today via either `ALTIMATE_VALIDATORS_ENABLED=1`
+(enforcement mode — failing validators block "done" with synthetic
+retries) or `ALTIMATE_VALIDATORS_SHADOW=1` (telemetry-only mode — runs
+without blocking, useful for measuring "would have caught" rates). When
+neither flag is set the dispatch path is completely skipped and there
+is zero overhead. See the [Validators page](../validators.md) for the
+full reference, env var catalogue, performance characteristics, and
+the phased rollout plan.

docs/docs/data-engineering/validators.md

@@ -117,7 +117,7 @@ minute of wall time.
 ## Telemetry
 
 When validators run (either mode), they emit one
-`validator_check` event per applied validator plus a session rollup:
+`validator_check` event per applied validator:
 
 ```json
 {
@@ -156,11 +156,12 @@ When a validator runs, it returns:
   fixHint?: string     // the body injected into the synthetic user turn
   details: {
     models_touched: number
-    checked: number
     dbt_root: string | null
     session_id: string
-    concurrency_limit: number
     elapsed_ms: number
+    // present only when at least one model was touched:
+    checked?: number
+    concurrency_limit?: number
     // validator-specific extras:
     // dbt-tests-pass:
     passed?: number