Add benchmarks/behavior-stability/ with a single local scenario:

- constraint-persistence: verifies that ContextManager.fold()
  preserves pinned constraints (HIGH PRIORITY, User memory, Project
  memory) from the system prompt into the fold summary.

This is a deterministic, fast test that exercises the core fix in
PR esengine#1515 without requiring API calls.

Usage:
  npx tsx benchmarks/behavior-stability/runner.ts --local

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

Add benchmarks/behavior-stability/ with a single local scenario:

- constraint-persistence: verifies that ContextManager.fold()
  preserves pinned constraints (HIGH PRIORITY, User memory, Project
  memory) from the system prompt into the fold summary.

This is a deterministic, fast test that exercises the core fix in
PR esengine#1515 without requiring API calls.

Usage:
  npx tsx benchmarks/behavior-stability/runner.ts --local

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

)

Small follow-up to #1515.

`extractPinnedConstraints` used `.match()` without the `g` flag, which only returns the first capture. The system prompt can carry multiple blocks under the same prefix:

- `# User memory — global` and `# User memory — this project` both exist when a user has both (`src/memory/user.ts` emits them as separate sections).
- Multiple `# Project memory (filename)` sections can show up when a project carries per-subdir memory.

Single-match meant the second one was silently dropped from the fold-tail constraint reinforcement.

Combined the three prefixes into one regex and switched to `matchAll` so every pinned block makes it into the fold summary.

…se (#1565)

* chore(release): 0.49.0 — static-history TUI, queued steers, Bing default, lifecycle plans

Headline themes:
- TUI: Static-history renderer is the only path; virtual-viewport layers removed (#1529 stages 1-4)
- Chat: queued mid-turn steer handling so input mid-render doesn't drop or fight the live frame (#1501)
- Web search: default switches to Bing; dashboard engine switcher; Mojeek dropped (#1558)
- Plans: lifecycle evidence summaries surface why a plan is ready to accept (#1500)
- Desktop: native OS notifications for approvals + completion (#1519)
- i18n: CLI command output (/mcp /sessions /prune /theme) + approval-prompt labels translated (#1524, #1560)
- Security: SSRF block in web_fetch (#1544), edit-snapshot path containment (#1454), shell redirect sandbox (#1457), Task integrity guardrail (#1516)
- Tools: per-turn dispatch-rate limit (#1356); run_command discourages shell-based edits (#1514)
- Client: DeepSeek 429 → concurrency-limit hint (#1526); timeoutMs honored with AbortSignal (#1535); --no-proxy opt-out for direct route (#1507)
- Files: read/edit/restore preserves source encoding (GB18030 / UTF-8 BOM) (#1518)
- Context: pinned constraints survive folds + full tail capture (#1515, #1552)
- Refactor: lifecycle risk policy extracted into its own module (#1557)

See CHANGELOG for the full list.

* fix(context): align fold summary prefix with main agent for cache reuse

The summarizer call was sending a bespoke "You compress conversation
history" system prompt and no tools, guaranteeing a 0% cache hit
against the main agent's just-cached prefix. Reshape the request so
system + tools + head bytes mirror the live agent's last call — the
only novel bytes are the trailing summarize instruction.

Skill-pin handling now collects bodies read-only instead of stubbing
mid-head, so the cache prefix stays unbroken. The summarize
instruction names pinned skills so the model knows not to paraphrase
their bodies (which we append verbatim regardless).

Measured on a real session at 48.7K prompt tokens:
  OLD shape: 0.0% cache hit  → $0.145 per fold
  NEW shape: 99.6% cache hit → $0.015 per fold
  saving: 89.6% per fold

* tools: add fold-cache shape + live benchmarks

bench-fold-cache-shape.mjs replays real session jsonls, simulates
OLD vs NEW summary-call shapes at the fold point, and reports
byte-level shared-prefix with the main agent's preceding request.
Pure local — no API required.

bench-fold-cache-live.mjs sends one priming + two summary calls to
DeepSeek and reports prompt_cache_hit_tokens / cost for each shape.
Used to confirm the shape change actually translates to API-side
cache hits.

---------

Co-authored-by: reasonix <reasonix@deepseek.com>