Clue Guide
As a detective, you read clues on each character‘s card to deduce who is a Suspect and who is Innocent. Griductive uses 19 distinct clue types. Click any one to read how it works.
Direct Classification
A specific character is directly confirmed as either a Suspect or Innocent — no deduction required. The constraint is simply: that one cell must equal the stated status. Use it immediately to lock in that cell, then feed the result into any other clues that reference the same row, column, or neighborhood.
Exact Count in Group
The clue states that exactly N cells in a named group — a row, column, edge ring, corner set, neighborhood, or directional slice — have the given status (Suspect or Innocent). The constraint is strict equality: the count must be exactly N, never more, never less. Once you have confirmed enough cells in that group, the rest are forced. Common mistake: treating this as a minimum when you should be eliminating the last unknown cell the moment you reach the stated count.
Total Grid Count
The total number of Suspects (or Innocents) across every cell in the entire grid equals the stated number. Think of it as a global budget: every Suspect you place anywhere draws from the same pool, and every Innocent is implied by the remainder. It is most powerful when combined with row or column count clues — if you know the total and the individual group counts, you can often compute the count for whichever group has no explicit clue.
Minimum Count in Group
At least N cells in the named group are Suspects (or Innocents). This is a lower-bound constraint (≥ N), so it can never be violated by having more than N. It rules out the possibility that the group contains fewer than N. By itself it cannot force an exact count, so pair it with an upper-bound source such as a total-grid count or an exact count in a complementary group. A frequent error is failing to act on this clue when it directly conflicts with an already-deduced set of Innocents in the same group.
Equal Counts in Two Groups
Two named groups — such as two rows or two columns — contain exactly the same number of Suspects (or Innocents). The constraint equates their counts but says nothing about what that shared count is. Resolve one group first using other clues, and the second group's count becomes known for free. Note that the two groups may overlap (e.g., a row and a column share one cell); if so, the shared cell counts toward both sides of the equation simultaneously.
Odd Count in Group
The number of Suspects (or Innocents) in the named group is odd — meaning 1, 3, 5, and so on. All even values (0, 2, 4…) are ruled out. In a 3-cell group such as a row, that means the count is either 1 or 3. In a larger group like the edge ring (8 cells in a 3×3), valid counts are 1, 3, 5, or 7. Combine with count bounds from other clues to narrow the odd value down to a single possibility. A common trap: players forget that 0 is even and therefore excluded.
More in One Group Than Another
The first named group contains strictly more Suspects (or Innocents) than the second named group — ties are not allowed. This creates an ordering between the two counts. To use it: enumerate the possible count pairs (a, b) for the two groups given whatever other constraints you know, then discard any pair where a ≤ b. Useful when combined with a total-grid count, because the two counts must sum to a fixed value while still satisfying the strict inequality.
Character Is One of N in Group
A named character is a Suspect (or Innocent), and their group — a row, column, or neighborhood — contains exactly N characters with that status. The clue combines two facts: (1) the named character definitely has the stated status, and (2) the group's total count is exactly N. That means N − 1 other characters in the same group also share that status. Use the confirmed character's status first, then apply the exact-count constraint to determine the remaining unknowns. Important: the anchor card carrying this clue is a third-party observer — it is not the named character.
Group Has the Most
A specific group — such as a row, column, or a cell's 8-neighborhood — contains strictly more Suspects (or Innocents) than every other group of the same kind — it is the unique maximum. If any other group ties or exceeds it, the clue is violated. To use it: identify the candidate with the highest possible count and verify that it exceeds all others; conversely, rule out any configuration where a second group reaches the same peak. Most useful when the gap between the leader and its peers is exactly 1, because that tightly constrains every other group's count.
Two Suspects Are Connected
The named group contains exactly 2 Suspects (or Innocents), and those two are connected to each other — sharing an edge (4-connectivity, not diagonal). The constraint has two parts: the group total is exactly 2, and the two must be directly adjacent. In a linear group like a row or column, this rules out any pair of cells with an unoccupied cell between them (i.e., the two must be next to each other). Common mistake: confusing adjacency with proximity — two cells with one innocent cell between them are NOT connected under this clue, and diagonal neighbors do not satisfy it either.
Only One Group Has Exactly N
Among all groups of the same kind — all rows, or all columns — exactly one of them has a count of exactly N Suspects (or Innocents). Every other group must have a count different from N (either higher or lower). This is a global uniqueness constraint. It is violated as soon as two groups simultaneously reach N; therefore any deduction that would create a second group with count N must be rejected. Work by first identifying which group is the unique N-holder, then force all others away from N.
Two Groups Share Exactly N
Two named groups overlap at one or more cells, and exactly N of those shared cells are Innocents. For a row and a column, the intersection is a single cell — so this clue simply pins whether that one cell is Innocent or Suspect. For larger overlapping groups (e.g., a row and a neighborhood), the intersection can contain several cells, and the count of Innocents among them must equal N exactly. Use individual group counts alongside this shared count to constrain the cells that belong to only one of the two groups.
All Suspects Form a Chain
All Suspects (or Innocents) within the named group form a single connected component under 4-neighbor (orthogonal) adjacency. In a linear group such as a row or column, this simply means the Suspects occupy a contiguous block with no gaps. The clue implicitly requires at least 2 Suspects in the group (a single cell is trivially connected and not informative enough to generate). When you know both the count and the connectivity requirement, the valid placements are only the contiguous sub-segments of that length — far fewer than the unrestricted count would suggest.
This Specific Group Has Exactly N
This specific group has exactly N Suspects (or Innocents), and no other group of the same kind shares that count. The clue asserts two things simultaneously: (1) the count in this group is exactly N, and (2) every other group of the same kind has a count that is not N. Use both halves together: pin the count for this group, then eliminate any assignment where a second group would also reach N. Compared to "Only One Group Has Exactly N", this variant identifies which specific group is the unique holder, making the first half immediately actionable.
Max Suspect Neighbors per Person
For every cell in the named group, the number of Suspects (or Innocents) in that cell's 8-neighborhood (king-move distance 1, including diagonals) is at most N. The constraint applies uniformly to every member of the group, not just one person. Note that neighbors are counted globally — they are not restricted to cells inside the group itself. This makes the clue sensitive to cells outside the group boundary. Use it to cap how many Suspects can cluster in the vicinity of any single cell in the group, which is especially useful for ruling out dense concentrations.
Two Groups Share an Odd Number
The number of Innocents in the overlap between two named groups is odd (1, 3, 5…). If the intersection is a single cell — as it always is when a row and a column cross — this reduces to: that cell is an Innocent (since 1 is odd and 0 is even). For larger overlaps, valid counts are any odd values from 1 up to the size of the intersection. An empty intersection makes the clue unsatisfiable. The power of this clue comes from parity reasoning: combine it with individual group totals to determine whether the overlap cell-set must contribute an odd or even number to each group's budget.
Group Shares K of Its N Suspects with Another
The first group has exactly N Suspects (or Innocents) in total, and exactly K of those N are cells that also belong to the second group. This encodes two simultaneous constraints: (1) an exact count for the first group, and (2) an exact count for the intersection of the two groups. The remaining N − K Suspects from the first group must lie outside the intersection, which constrains the cells belonging only to the first group. Reason from both the total count and the intersection count together — neither alone is sufficient.
Exactly One Person Has N Suspect Neighbors
Among all cells in the named group, exactly one cell has exactly N Suspects (or Innocents) in its 8-neighborhood. All other cells in the group must have a neighbor count different from N. Neighbors are counted globally across the whole grid, not only within the group. This requires testing each candidate cell: which one can have exactly N Suspect neighbors while every other group member simultaneously has a different count? The constraint is violated as soon as two cells in the group both achieve a neighbor count of N.
Every Group Has at Least N
Every group of the stated kind — every row, every column, or every cell's neighborhood — contains at least N Suspects (or Innocents). The constraint is applied independently to each group simultaneously; if any single group falls below N, the clue is violated. This sweeping lower bound becomes powerful when combined with a global total count: if you know every row has at least N Suspects and there are R rows, the total must be at least R × N. From there, an exact total count or per-group upper bounds can often force the entire distribution. Most difficult when N is high relative to group size, since it leaves little room for Innocents.