strata-agent-memory 0.2.5__tar.gz → 0.2.7__tar.gz

{strata_agent_memory-0.2.5/strata_agent_memory.egg-info → strata_agent_memory-0.2.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: strata-agent-memory
-Version: 0.2.5
+Version: 0.2.7
 Summary: Portable, auditable, cross-vendor memory for LLM agents: evidence-bound claims, trust tiers, and compiled context views.
 Author: Strata Intelligence
 License-Expression: Apache-2.0
@@ -75,7 +75,7 @@ pip install strata-agent-memory     # the CLI is `agent-memory`; or, from a chec
 cd your-project && agent-memory init
 
 agent-memory run -- pytest -q                     # -> {"id": "ev_…", "kind": "test", "passed": 42, ...}
-agent-memory claim outcome "auth tests pass" -e ev_… --commit
+agent-memory claim outcome "auth tests pass" -e ev_…   # inside a git checkout it binds HEAD too
 agent-memory claim outcome "deployed" # -> rejected: outcome claims require at least one receipt
 
 agent-memory claim fact "sessions renew via renewSession()" --source src/auth.py
@@ -91,6 +91,7 @@ agent-memory claim fact "w3lib.get_meta_refresh takes baseurl" --source w3lib/ht
 agent-memory compile --role worker --task "extend auth"    # a fork-mode brief; reports what memory holds nothing on
 agent-memory compile --role verifier --contract ctr_… --round 2 --lens security   # isolated review brief with a lens
 agent-memory audit                                          # tiers, static-only outcomes, taint chains, verifier lineage matrix
+agent-memory audit --since 2026-09-21 --until 2026-09-28    # and the measurement section for that week (default: the last seven days)
 agent-memory reliability                                    # how often claims like these were refuted, with n and an interval; fitted staleness
 AGENT_MEMORY_HARNESS=ollama agent-memory recheck --map /Users/craig/src=/home/chris/src   # the nightly lane, from another machine
 agent-memory serve                                          # HTTP API, SSE and remote MCP on loopback; add --tokens FILE

{strata_agent_memory-0.2.5 → strata_agent_memory-0.2.7}/README.md

@@ -51,7 +51,7 @@ pip install strata-agent-memory     # the CLI is `agent-memory`; or, from a chec
 cd your-project && agent-memory init
 
 agent-memory run -- pytest -q                     # -> {"id": "ev_…", "kind": "test", "passed": 42, ...}
-agent-memory claim outcome "auth tests pass" -e ev_… --commit
+agent-memory claim outcome "auth tests pass" -e ev_…   # inside a git checkout it binds HEAD too
 agent-memory claim outcome "deployed" # -> rejected: outcome claims require at least one receipt
 
 agent-memory claim fact "sessions renew via renewSession()" --source src/auth.py
@@ -67,6 +67,7 @@ agent-memory claim fact "w3lib.get_meta_refresh takes baseurl" --source w3lib/ht
 agent-memory compile --role worker --task "extend auth"    # a fork-mode brief; reports what memory holds nothing on
 agent-memory compile --role verifier --contract ctr_… --round 2 --lens security   # isolated review brief with a lens
 agent-memory audit                                          # tiers, static-only outcomes, taint chains, verifier lineage matrix
+agent-memory audit --since 2026-09-21 --until 2026-09-28    # and the measurement section for that week (default: the last seven days)
 agent-memory reliability                                    # how often claims like these were refuted, with n and an interval; fitted staleness
 AGENT_MEMORY_HARNESS=ollama agent-memory recheck --map /Users/craig/src=/home/chris/src   # the nightly lane, from another machine
 agent-memory serve                                          # HTTP API, SSE and remote MCP on loopback; add --tokens FILE

{strata_agent_memory-0.2.5 → strata_agent_memory-0.2.7}/agent_memory/__init__.py

@@ -9,4 +9,4 @@ see, what did it claim, and can the proof be re-run.
 The core is dependency-free (Python 3.10+ standard library only).
 """
 
-__version__ = "0.2.5"
+__version__ = "0.2.7"

{strata_agent_memory-0.2.5 → strata_agent_memory-0.2.7}/agent_memory/audit.py

@@ -27,13 +27,14 @@ from collections import Counter, defaultdict
 from typing import Dict, List, Optional
 
 from . import evidence as ev
+from . import measure as measure_mod
 from . import reliability as rel
 from . import supersession as supersession_mod
 from . import roster as roster_mod
 from . import signing
 from .contracts import status_of
 from .ledger import UNTRUSTED_POLICY
-from .records import is_redacted, receipt_visible_to
+from .records import is_assessment, is_redacted, receipt_visible_to
 from .trust import TIERS, assess, lineage_differs, summarize
 
 # Anything a compiled view would have to render on its own line. The renderer collapses
@@ -190,12 +191,15 @@ def _roster_sections(ledger, rep: Dict, by_id: Dict[str, dict], reader_user: Opt
 
 
 def report(ledger, recheck: bool = False, cwd: Optional[str] = None, timeout: Optional[float] = 300,
-           path_map=None, reader_user: Optional[str] = None) -> dict:
+           path_map=None, reader_user: Optional[str] = None, since: Optional[str] = None,
+           until: Optional[str] = None) -> dict:
     """The audit, for one reader. ``reader_user`` is who is asking: a user-scoped claim is
     one person's private memory and is listed for its owner and for nobody else, the same
     rule ``receipt_visible_to`` applies everywhere a record is shown (verification review
     F1: the audit listed every user's private statements to any token). A local audit
-    with no reader named sees everything, as the local ledger's owner does."""
+    with no reader named sees everything, as the local ledger's owner does. ``since`` and
+    ``until`` are the measurement section's window (measure.window; ValueError when it is
+    empty or a bound does not parse)."""
     assessment = assess(ledger)
     claims = ledger.claims()
     if reader_user is not None:
@@ -214,10 +218,14 @@ def report(ledger, recheck: bool = False, cwd: Optional[str] = None, timeout: Op
         "unrechecked_outcomes": [],
         "failing_receipt_outcomes": [],
         "static_only_outcomes": [],
+        "dirty_bound_outcomes": [],
         "assurance_overridden": [],
         "same_vendor_only": [],
         "self_verified_by_author": [],
         "same_session_different_harness": [],
+        # D6 (ctr_67b26eae6bff4b63): another identity's reading of a claim, asked for; listed
+        # here and nowhere a verdict is counted
+        "assessments": [],
         "formatted_statements": [],
         "redactions": {"count": 0, "ids": []},
         # absent and empty mean different things (an absent key is the starter default
@@ -298,6 +306,12 @@ def report(ledger, recheck: bool = False, cwd: Optional[str] = None, timeout: Op
             if backed in ("static", "attested"):
                 # a clean static gate is not evidence that the thing runs
                 rep["static_only_outcomes"].append({**entry, "assurance": backed})
+            # the writer's own word that its receipts ran on a tree no sha reproduces: no
+            # lane checks those out, so a practice of it is a PM's question per author (S3)
+            heads = [item.get("sha") for item in b.get("evidence") or []
+                     if isinstance(item, dict) and item.get("kind") == "commit" and item.get("dirty")]
+            if heads:
+                rep["dirty_bound_outcomes"].append({**entry, "user": a.get("user"), "head": heads[0]})
         # a measure, not a tier: every claim whose only verifiers are agents of its own
         # vendor, whether the ledger's policy lets that reach `verified` or not
         if info["verified_by"] and not info.get("cross_vendor") and not info.get("self_verified_by_author") \
@@ -351,6 +365,17 @@ def report(ledger, recheck: bool = False, cwd: Optional[str] = None, timeout: Op
         claim = by_id.get(v["body"].get("claim"))
         if claim is None:
             continue
+        if is_assessment(v):  # a reading, not a verdict: its own column, never the matrix
+            rep["assessments"].append({
+                "id": v["id"], "claim": claim["id"], "verdict": v["body"]["verdict"],
+                "by": v["author"].get("harness"), "vendor": v["author"].get("vendor"),
+                "user": v["author"].get("user"), "ts": v["ts"],
+                "lineage": ("another lineage" if lineage_differs(claim["author"], v["author"])
+                            else "same lineage: never counted"),
+                "note": str(v["body"].get("note") or "")[:160], "cited": list(v["body"].get("cited") or []),
+                "statement": claim["body"]["statement"][:120],
+            })
+            continue
         verdict = v["body"]["verdict"]
         key = f"{claim['author'].get('vendor')}->{v['author'].get('vendor')}"
         cell = rep["lineage_matrix"].setdefault(key, {"verified": 0, "refuted": 0, "inconclusive": 0})
@@ -463,6 +488,11 @@ def report(ledger, recheck: bool = False, cwd: Optional[str] = None, timeout: Op
                 rep["recheck"].append(entry)
     # what the reliability measurements leave out, and why (the PE's ruling, item 4)
     rep["not_measured"] = {"superseded": len(rep["superseded"]), "artefact": len(rep["withdrawn"])}
+    # the vision's numbers for a window, recomputable by anyone with the shards
+    # (ctr_13190a200fbbd682; docs/BENCHMARK-PLAN.md section 3)
+    rep["measurement"] = measure_mod.measurement(
+        ledger, since, until, sees=lambda rec: _reader_sees(ledger, rec, by_id, reader_user),
+        assessment=assessment)
     return rep
 
 
@@ -516,6 +546,9 @@ def format_report(rep: dict) -> str:
             lambda r: f"{r['id']} [{r['by']}] {r['statement']} -- receipts: {', '.join(r['receipts'])}")
     section("Outcome claims backed only by static evidence (static-pass/dynamic-fail risk)",
             rep["static_only_outcomes"], lambda r: f"{r['id']} [{r['by']}, {r['assurance']}] {r['statement']}")
+    section("Outcome claims bound to a dirty tree (their commit is not what ran; no lane re-runs them at it)",
+            rep.get("dirty_bound_outcomes", []),
+            lambda r: f"{r['id']} [{r['by']}, {r['user']}] at {str(r['head'])[:12]} {r['statement']}")
     section("Receipts whose assurance was declared above the computed default", rep["assurance_overridden"],
             lambda r: f"{r['id']} [{r['by']}] declared {r['declared']}, computed {r['computed']}: {r['what']}"
                       + (f" (cited by {', '.join(r['claims'])})" if r["claims"] else " (cited by no claim)"))
@@ -527,6 +560,9 @@ def format_report(rep: dict) -> str:
             rep.get("same_session_different_harness", []),
             lambda r: f"{r['id']} [{r['by']} -> {r['verified_by']}, {r['user']} session {r['session']}] "
                       f"{r['statement']}")
+    section("Assessments (a reading on request, never a check: they move no tier)", rep.get("assessments", []),
+            lambda r: f"{r['id']} on {r['claim']} [{r['verdict']} by {r['by']}, {r['vendor']}, {r['user']}; "
+                      f"{r['lineage']}] {r['note']}" + (f" (cites {', '.join(r['cited'])})" if r["cited"] else ""))
     section("Statements shaped like a document (injection risk)", rep.get("formatted_statements", []),
             lambda r: f"{r['id']} [{r['by']}] {r['field']} starts a line with: {r['line']}")
     section("Refuted", rep["refuted"], lambda r: f"{r['id']} {r['statement']} -- {'; '.join(r['reasons'])} (dependents: {r['dependents']})")
@@ -617,6 +653,9 @@ def format_report(rep: dict) -> str:
                                         "same_lineage_agreement": rep["same_lineage_agreement"],
                                         "staleness": rep["staleness"],
                                         "judgment_calibration": rep["judgment_calibration"]}).rstrip("\n"))
+    if rep.get("measurement"):  # a replica older than this section sends none
+        lines.extend(measure_mod.format_noise(rep["measurement"]))
+        lines.extend(measure_mod.format_measurement(rep["measurement"]))
     if rep["recheck"]:
         section("Evidence re-check", rep["recheck"],
                 lambda r: f"{r['claim']} {r['evidence']}: {r['outcome']} -- {r['detail']}"