fix(knowledge_graph): use named column access instead of hardcoded integer indices

[christauff]

Problem

query_entity(), query_relationship(), and timeline() access JOIN query results using hardcoded integer column indices (row[2], row[4], row[5], row[6], row[7], row[10], row[11]).

These indices are derived from the position of columns in SELECT t.*, ... — which means any schema migration (adding a column, reordering columns, changing a JOIN) silently corrupts results without raising an exception. The bug would surface as wrong data returned from graph queries.

Example from query_entity() outgoing direction:

results.append({
    "subject": name,
    "predicate": row[2],   # triples.predicate — but fragile
    "object": row[10],     # obj_name from JOIN — breaks if schema changes
    ...
})

Fix

Use cursor.description to build a column name → value dict after every execute() call. All field access now uses named keys:

cur = conn.execute(query, params)
cols = [d[0] for d in cur.description]
for raw in cur.fetchall():
    row = dict(zip(cols, raw))
    results.append({
        "predicate": row["predicate"],
        "object": row["obj_name"],
        ...
    })

This is schema-safe: the column name contract is expressed in the SQL AS alias, not in a magic integer offset.

Test plan

• kg.query_entity("Alice", direction="both") returns correct subject/predicate/object names
• kg.query_relationship("child_of") returns correct sub_name/obj_name
• kg.timeline() returns facts in chronological order with correct field values
• Add a column to triples table schema and verify queries still return correct data

🤖 Generated with Claude Code

[bensig]

Clean fix. CI lint is failing — can you run ruff check . && ruff format --check . and fix any issues?

[web3guru888]

🔧 Review of #89 — fix(knowledge_graph): use named column access instead of hardcoded integer indices

Scope: +180/−41 · 2 file(s) · touches core

• ⚠️ mempalace/knowledge_graph.py (modified: +50/−41)
• tests/test_knowledge_graph.py (added: +130/−0)

Technical Analysis

• 🕸️ Knowledge graph changes — verify temporal triple ordering and connection lifecycle

Issues

• ⚠️ Touches mempalace/knowledge_graph.py — Core KG — threading and concurrency sensitive

Suggestions

• Magic number(s) 2015, 2025 — consider extracting to named constant(s)

Strengths

• ✅ Includes test coverage

🟡 Needs attention — touches guarded files and has items to address.

---

_{🏛️ Reviewed by MemPalace-AGI · Autonomous research system with perfect memory · Showcase: Truth Palace of Atlantis}

[igorls]

Hi, thanks for the contribution.

This PR has merge conflicts with develop, and the branch has not been updated in over 7 days, which puts it before our most recent release. The conflicts are likely against work that landed in that release.

Could you rebase onto develop so we can take another look?

If this change is no longer relevant, feel free to close the PR.

(This message is part of a periodic backlog pass, sent to all open PRs that match this state.)