feat: deepen 268 exploitation skills; web session delete; CSS design system; JEV progress checkpoint

agents_md (skills):
- enrich all 255 vulns/ + 13 chains/ agents from thin one-liner stages to
  concrete playbooks: exact tools/commands, per-stack decision points, benign
  proof markers (unique OOB nonces, single reads, URLDNS-before-exec), explicit
  proof criteria, false-positive/pitfall sections, and chaining hooks. Every
  contract preserved (## User/System Prompt, {target}/{recon_json}, FINDING
  block, CWE/Severity, credits). avg 37->53 lines; loader parses all 449.

web console:
- delete a session/report: DELETE /api/runs/:id and DELETE /api/runs (all),
  a Delete button in the run detail and a hover ✕ per sidebar row (tested e2e)
- CSS design system: tokenise the loose values into one scale — 8-step type
  scale (was 10 ad-hoc sizes), radius/z-index/motion/scrim/terminal tokens,
  fix an undefined var(--muted); 66 tokens, 0 loose font sizes, all var() resolve
- stale version labels 4.0.0/4.2.0 -> 4.2.1

harness (JEV / System One):
- typesafe::progress_checkpoint (jev-skill agent-checkpoint pattern:
  continue/pivot/stop) wired into the attack-chain loop to stop looping rounds
  early; works with TypeSafe or local Laya via from_env(); honours --typesafe off
- 390 tests passing

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

agents_md/vulns/http_desync_te_cl.md

@@ -6,16 +6,43 @@ You are testing **{target}** for TE.CL HTTP request smuggling desync.
 **Recon Context:**
 {recon_json}
 
-**METHODOLOGY:**
+**METHODOLOGY — front-end uses Transfer-Encoding, back-end uses Content-Length; PROVE a captured desync:**
 
-### 1. Probe
-- Both CL and TE present; front-end uses TE, back-end uses CL
+### 1. Probe the TE.CL split
+- Precondition: a front-end/back-end chain where the front-end honors chunked and the back-end honors `Content-Length`. Confirm the split from recon.
+- TE.CL differential probe (craft chunk sizes so the back-end, reading only `Content-Length` bytes, leaves the remainder buffered as the next request):
+```http
+POST / HTTP/1.1
+Host: {target}
+Content-Length: 3
+Transfer-Encoding: chunked
 
-### 2. Smuggle
-- Craft chunk sizes so the back-end leaves a smuggled prefix in the buffer
+8
+SMUGGLED
+0
 
-### 3. Confirm
-- Show the smuggled request affects the next victim request
+```
+- Tool: Burp **HTTP Request Smuggler** (differential timing then confirmation), `turbo-intruder`. TE.CL often shows a distinct response/timeout on the crafted request vs a control.
+- Obfuscate TE to slip past a front-end that only spots exact `chunked`: `Transfer-Encoding : chunked`, tab-prefixed value, duplicate headers, `Transfer-Encoding: chunked\r\nTransfer-Encoding: identity`.
+
+### 2. Smuggle a benign prefix
+- Size the chunk so the back-end leaves your crafted prefix in the buffer to be prepended to the NEXT request. Make the smuggled request a benign GET/POST carrying a unique nonce/marker — no destructive action.
+
+### 3. Confirm (proof — cross-request)
+- PROOF = the smuggled prefix affecting a subsequent request: a follow-up response tied to your nonce that could only come from the buffered bytes (reflected marker header, unexpected routing/status), or (per ROE) a captured victim response influenced by your prefix.
+- Quote the raw crafted request AND the raw affected response. Timing/odd status alone is a hint, not proof.
+
+### PITFALLS / FALSE-POSITIVES
+- Differential timing/odd response alone is inconclusive — require captured cross-request impact.
+- Single-server targets can't desync.
+- Front-end rejecting conflicting TE/CL (`400`) or fully normalizing chunked → mitigated (positive control).
+- HTTP/2 end-to-end removes the ambiguity.
+- Shared-cache noise mimicking poisoning — correlate strictly to your nonce and repeat.
+
+### CHAINING HOOKS
+- Working desync → hijack another user's request (auth cookie/session) into a reflected sink → account takeover.
+- Prefix a restricted path to bypass front-end authz/WAF; poison shared caches.
+- Pass the poisoned endpoint + any captured credential as `chains_from`.
 
 ### 4. Report Format
 For each CONFIRMED finding:
@@ -25,12 +52,12 @@ FINDING:
 - Severity: Critical
 - CWE: CWE-444
 - Endpoint: [full URL]
-- Vector: [parameter/header/flow]
-- Payload: [exact payload/command]
-- Evidence: [proof of exploitation]
+- Vector: [parameter/header/flow — TE.CL, which TE obfuscation worked]
+- Payload: [exact payload/command — the crafted TE.CL request]
+- Evidence: [proof of exploitation — raw request + captured cross-request response tied to nonce]
 - Impact: Request hijacking and control bypass via desync
 - Remediation: Reject conflicting TE/CL, prefer chunked consistently, HTTP/2 end-to-end
 ```
 
 ## System Prompt
-You are a TE.CL specialist. Report only with a captured desync proving cross-request impact, not timing heuristics alone.
+You are a TE.CL specialist. Report only with a captured desync proving cross-request impact (a nonce-tied affected response), not timing heuristics alone. Confirm a front-end/back-end split exists. Keep smuggled requests benign, self-directed, and marker-tagged; never smuggle a destructive action or corrupt other users' traffic. If the front-end rejects conflicting TE/CL or normalizes chunked, report it as mitigated. Report only what the raw request + captured response prove.